CN115544037A - Transaction execution method and distributed database system - Google Patents

Abstract

This application provides a transaction execution method and a distributed database system, which relate to the field of cloud computing. In this embodiment, the coordination node CN and the data node DN use the first start timestamp and timestamp list of the target transaction provided by the timing node TSO , modify the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction; execute the target transaction according to the globally consistent active transaction information. This embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the time stamp assigned by the TSO to achieve global transaction consistency.

Description

Transaction execution method and distributed database system

technical field

The present application relates to the technical field of cloud computing, and in particular to a transaction execution method and a distributed database system.

Background technique

In mainstream stand-alone relational databases, Multi-Version Concurrency Control (MVCC) is generally used as a concurrency control technology to satisfy transaction consistency. Provide transaction isolation for each database session through multi-version concurrency control. With the increasing data, stand-alone databases can no longer meet the demand, and distributed databases have developed rapidly. Distributed databases can expand storage and throughput by expanding nodes.

However, under the distributed architecture, data distribution and transaction processing will involve different physical machines. The original stand-alone MVCC technology cannot guarantee the consistency of global transactions, resulting in incorrect execution of transactions.

Contents of the invention

The embodiment of the present application provides a transaction execution method and a distributed database system, so as to realize the consistency of the global transaction in the distributed database and ensure the correct execution of the transaction.

In the first aspect, the embodiment of the present application provides a transaction execution method, the method is applied to the coordination node CN in the database, and the method includes:

Receive the execution request of the target transaction and obtain the local active transaction information; the transactions in the local active transaction information include the uncommitted transactions;

Obtain the first start timestamp and the timestamp list of the target transaction from the timing node TSO in the database, the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the submission before the first start timestamp;

According to the first start timestamp and the timestamp list, modify the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction;

According to the global consistent active transaction information, the target transaction is executed to achieve the consistency of the global transaction.

In the second aspect, the embodiment of the present application provides a transaction execution method, the method is applied to the data node DN in the database, and the method includes:

Receive the first start time stamp and time stamp list of the target transaction sent by CN or TSO, the time stamp list includes the second start time stamp of the transaction that has completed the preparation and has not completed the commit before the first start time stamp;

Obtain local active transaction information; transactions in local active transaction information include uncommitted transactions;

According to the first start timestamp and the timestamp list, modify the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction;

According to the global consistent active transaction information, the target transaction is executed to achieve the consistency of the global transaction.

In the third aspect, the embodiment of the present application provides a transaction execution method, the method is applied to the timing node TSO in the database, and the method includes:

Receive the start timestamp acquisition request of the target transaction sent by CN or DN;

Send the first start timestamp and timestamp list of the target transaction to CN or DN, so that CN or DN can modify the local active transaction information according to the first start timestamp and timestamp list, and obtain the globally consistent active transaction corresponding to the target transaction Information; the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp; the global consistency active transaction information is used for CN or DN to execute the target transaction to achieve the consistency of the global transaction sex.

In a fourth aspect, the embodiment of the present application provides a transaction execution method, the method including:

The timing node TSO receives the start timestamp acquisition request of the target transaction sent by the coordination node CN or data node DN;

TSO sends the first start timestamp and timestamp list of the target transaction to the CN or DN; the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp;

CN or DN modifies the local active transaction information according to the first start timestamp and the timestamp list, and obtains the globally consistent active transaction information corresponding to the target transaction;

CN or DN executes the target transaction according to the globally consistent active transaction information to achieve global transaction consistency.

In a fifth aspect, an embodiment of the present application provides a distributed database system, which includes the coordination node CN, data node DN, and timing node TSO in any of the above embodiments.

In a sixth aspect, the embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory, and the processor implements any one of the methods above when executing the computer program.

In a seventh aspect, the embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, any one of the methods above is implemented.

Compared with the prior art, the present application has the following advantages:

The embodiment of the present application provides a transaction execution method and a distributed database system. The coordination node CN and the data node DN use the first start timestamp and timestamp list of the target transaction provided by the timing node TSO to modify the local active transaction information to obtain The globally consistent active transaction information corresponding to the target transaction; execute the target transaction according to the globally consistent active transaction information. In this embodiment, the TSO and local active transaction information are integrated, and the local active transaction information is modified according to the time stamp assigned by the TSO to achieve global transaction consistency. Through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

The above description is only an overview of the technical solution of the present application. In order to understand the technical means of the present application more clearly, it can be implemented according to the contents of the description, and in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, Specific embodiments of the present application are enumerated below.

Description of drawings

In the drawings, unless otherwise specified, the same reference numerals designate the same or similar parts or elements throughout the several drawings. The drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some implementations according to the application, and should not be considered as limiting the scope of the application.

FIG. 1 is a schematic diagram of a scenario of a transaction execution method provided by the present application;

FIG. 2 is a flowchart of a transaction execution method according to an embodiment of the present application;

FIG. 3 is a flowchart of a transaction execution method according to an embodiment of the present application;

FIG. 4 is a flowchart of a transaction execution method according to an embodiment of the present application;

FIG. 5 is an interaction diagram of a distributed transaction execution process at an RR/SSI isolation level according to an embodiment of the present application;

FIG. 6 is an interaction diagram of a distributed transaction execution process at an RC isolation level according to an embodiment of the present application;

FIG. 7 is an interaction diagram of a non-distributed transaction execution process at the RR/SSI isolation level according to an embodiment of the present application;

FIG. 8 is an interaction diagram of a non-distributed transaction execution process at an RC isolation level according to an embodiment of the present application;

FIG. 9 is a structural block diagram of a transaction execution device according to an embodiment of the present application;

FIG. 10 is a structural block diagram of a transaction execution device according to an embodiment of the present application;

FIG. 11 is a structural block diagram of a transaction execution device according to an embodiment of the present application; and

FIG. 12 is a block diagram of an electronic device used to implement an embodiment of the present application.

detailed description

In the following, only some exemplary embodiments are briefly described. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present application. Accordingly, the drawings and descriptions are to be regarded as illustrative in nature, and not restrictive.

In order to facilitate understanding of the technical solutions of the embodiments of the present application, the related technologies of the embodiments of the present application are described below. The following related technologies may be optionally combined with the technical solutions of the embodiments of the present application as optional solutions, and all of them belong to the protection scope of the embodiments of the present application.

Fig. 1 is a schematic diagram of an exemplary application scenario for implementing the method of the embodiment of the present application. As shown in Figure 1, the database system includes a timing node (Timerstamp Oracle, TSO), a coordination node (Coordinator node, CN) 1, a coordination node 2, a data node (Data node, DN) 1 and a data node 2, wherein the coordination node There can be multiple and data nodes, which can be configured according to specific needs. Each node can be deployed in its corresponding device, and the timing node can be in the form of a component, which can be deployed in the same device as the coordination node or data node, or can be deployed separately. Multiple data tables can be stored in the coordinating node or data node, and each data table can be used to store one or more data items.

The coordinating node receives the execution request of the target transaction sent by the application, and obtains the local active transaction information; the transaction in the local active transaction information includes the uncommitted transaction; obtains the first start timestamp and timestamp list of the target transaction from the TSO, and the timestamp The list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp; according to the first start timestamp and the timestamp list, modify the local active transaction information to obtain the global consistency corresponding to the target transaction Active transaction information; according to the global consistency active transaction information, execute the target transaction to achieve the consistency of the global transaction. Wherein, the local active transaction information may be a snapshot of the local active transaction list, and the local active transaction list includes identification information of the local active transaction. The target transaction can be a distributed transaction or a non-distributed transaction. Transaction (Transaction) refers to a series of operations performed as a single logical unit of work, which can include multiple statements, either completely executed or not executed at all. By grouping a set of related operations into an all-or-nothing unit, error recovery is simplified and applications are made more reliable. For a logical unit of work to be a transaction, it needs to satisfy the properties of atomicity, consistency, isolation, and durability. Distributed Transaction means that transaction participants, transaction-supporting devices, resource devices, and transaction managers are located on different nodes of different distributed systems.

If the target transaction is a distributed transaction, the DN receives the first start timestamp and timestamp list of the target transaction sent by the CN, and if the target transaction is a non-distributed transaction, then the CN or DN receives the first start of the target transaction sent by the TSO Timestamp and list of timestamps. Wherein, the time stamp list includes the second start time stamp of the transactions that have been prepared but not yet committed before the first start time stamp. CN or DN obtains the local active transaction information; the transactions in the local active transaction information include uncommitted transactions; according to the first start timestamp and the timestamp list, modify the local active transaction information to obtain the globally consistent active transaction corresponding to the target transaction Information; according to the global consistent active transaction information, execute the target transaction to achieve the consistency of the global transaction. If the target transaction is a distributed transaction, after the preparation of the target transaction is completed, CN obtains the submission timestamp from TSO; sends a submission operation request to the data node DN in the database, and the submission operation request carries the submission timestamp. After submitting the successful notification message, execute the submit operation.

TSO receives the start timestamp acquisition request of the target transaction sent by CN or DN; sends the first start timestamp and timestamp list of the target transaction to CN or DN, so that CN or DN according to the first start timestamp and timestamp list, Modify the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction; the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp; the global consistency active transaction The information is used for CN or DN to execute the target transaction to achieve global transaction consistency.

In a related technology, the global transaction manager (GlobalTransaction Manager, GTM) of the global active transaction table centrally maintains the state of all transactions in the cluster, and generates snapshots of their active transaction tables for each transaction. Each node transaction judges data visibility through snapshot to maintain global transaction consistency. However, with the increase of nodes, the number of global active transactions increases, resulting in larger snapshot data, which in turn causes the central node network to become a bottleneck, restricting the scalability of the cluster, unable to expand more nodes, and limiting the distributed database cluster. scale.

In another related technology, TSO provides a monotonically increasing timestamp, and MVCC is implemented based on the timestamp. If the MVCC based on the active transaction table is used in the database kernel, the MVCC mechanism needs to be redeveloped, which is very intrusive to the kernel, difficult to develop, and not as good in compatibility and stability as the original MVCC mechanism. At the same time, in the timestamp-based MVCC scheme, When the write timestamp is less than the latest read timestamp, the transaction will be terminated (Abort), increasing concurrency conflicts.

In addition, in the timestamp-based MVCC scheme, TSO needs to assign a commit timestamp to each submitted transaction for visible consistency judgment, which reduces transaction execution efficiency. This embodiment integrates the MVCC of the active transaction table. When each non-distributed transaction is submitted, the transaction global identifier has been allocated locally. The non-distributed transaction submission does not need to obtain the submission timestamp. According to the local active transaction table, it can Complete the visible consistency judgment and improve the efficiency of transaction execution.

In this embodiment, the TSO and local active transaction information are integrated, and the local active transaction information is modified according to the time stamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

The embodiment of the present application provides a transaction execution method, as shown in Figure 2 is a flow chart of the transaction execution method in an embodiment of the present application, the method is applied to the coordination node CN in the database, the method includes:

Step S201, receiving an execution request of a target transaction, and acquiring local active transaction information; the transactions in the local active transaction information include uncommitted transactions.

Step S202, obtain the first start timestamp and the timestamp list of the target transaction from the timing node TSO in the database, the timestamp list includes the second start timestamp of the transaction that has completed the preparation but has not yet completed the commit before the first start timestamp .

Step S203, according to the first start timestamp and the timestamp list, modify the local active transaction information to obtain globally consistent active transaction information corresponding to the target transaction.

Step S204, execute the target transaction according to the globally consistent active transaction information, so as to realize the consistency of the global transaction.

Wherein, the coordinating node CN receives the execution request of the target transaction, and the execution request may be sent by an application program or in other forms. After the CN receives the target transaction execution request, it acquires the local active transaction information. The specific form of the local active transaction information is not limited, for example, it may be a snapshot of the local active transaction list. The local active transaction list includes the identification information of the active transaction.

Wherein, the target transaction may be a distributed transaction or a non-distributed transaction. Transaction (Transaction) refers to a series of operations performed as a single logical unit of work, which can include multiple statements, either completely executed or not executed at all. By grouping a set of related operations into an all-or-nothing unit, error recovery is simplified and applications are made more reliable. For a logical unit of work to be a transaction, it needs to satisfy the properties of atomicity, consistency, isolation, and durability. Distributed Transaction means that transaction participants, transaction-supporting devices, resource devices, and transaction managers are located on different nodes of different distributed systems. Distributed transactions can query data across nodes, and non-distributed transactions are transactions that complete data queries within a single node.

The status information of the transaction includes: in process (Inprocess), completed preparation (prepared), committing (committing), and completed committing (committed). The local active transaction information includes the identification information of the uncommitted transaction, and the uncommitted transaction includes the transaction whose state information is executing, completing preparation and committing.

CN sends the timestamp acquisition request of the target transaction to the timing node TSO, and receives the start timestamp and timestamp list of the target transaction sent by TSO. TSO provides global centralized timing services, provides start timestamps for each transaction in a linear increment manner, and provides commit timestamps for distributed transactions.

Exemplarily, the CN modifies the snapshot of the local active transaction list according to the start timestamp and the timestamp list sent by the TSO, obtains the snapshot of the globally consistent active transaction list corresponding to the target transaction, and executes according to the snapshot of the globally consistent active transaction list Target transaction to achieve data consistency when multiple transactions are executed concurrently in at least one node.

The transaction execution method provided in this embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the time stamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

Wherein, the specific implementation of step S203 is shown in the following embodiments:

In a possible implementation, the local active transaction information is modified according to the first start timestamp and the timestamp list to obtain globally consistent active transaction information corresponding to the target transaction, including:

According to the status information of the transaction, the first start timestamp and the timestamp list, the transaction is deleted and/or added to the local active transaction information, and the globally consistent active transaction information corresponding to the target transaction is obtained.

In practical applications, the CN obtains transaction status information from the shared cache interval, and the transaction status information includes: in-process, prepared, committed, and committed. According to the status information of the transaction, the timestamp in the timestamp list and the start timestamp of the target transaction, filter the transaction, delete the identification information of the filtered transaction in the snapshot of the local active transaction list, and/or filter out The identification information of the transaction is added to the snapshot of the local active transaction list, and the identification information of the transaction may be the global identification xid of the transaction. In this embodiment, by deleting and/or adding transactions in the local active transaction information, the globally consistent active transaction information corresponding to the target transaction is obtained, which is used for the execution of the target transaction. By modifying the local active transaction information, the stand-alone database kernel is retained The MVCC based on the active transaction table in the middle does not need the MVCC development based on the time stamp, which reduces the modification of the kernel, and avoids the concurrency conflict caused by the dislocation of the write time stamp and the read time stamp, and realizes high expansion and high compatibility of the database kernel and high stability.

Among them, the transaction to be deleted is determined in the local active transaction information, see the following embodiments for details:

In a possible implementation, according to the status information of the transaction, the first start timestamp and the timestamp list, the transaction is deleted from the local active transaction information, and the globally consistent active transaction information corresponding to the target transaction is obtained, including:

In the local active transaction information, perform at least one of the following operations to obtain the globally consistent active transaction information corresponding to the target transaction: delete the transaction whose status information is submitting and whose commit timestamp is less than the first start timestamp; the commit timestamp is from Obtained by TSO; delete the transaction whose status information is completed and whose start timestamp is in the timestamp list.

Exemplarily, traversing the status information of the transactions in the shared memory, the transactions deleted in the snapshot of the local active transaction list include at least one of the following: the status information of the transaction is submitting, and the commit timestamp is less than the start timestamp of the target transaction ; The status information of the transaction is ready for completion, and the start timestamp of the transaction is in the timestamp list.

In a possible implementation, the method further includes:

Make transactions deleted in the local active transaction information visible to the target transaction.

In practical applications, pre-visibility settings are performed for transactions deleted from the snapshot of the local active transaction list, because deletion from the snapshot means that the transaction is visible to the target transaction, but because it has not yet completed the commit, according to global consistency Judgment criteria, the data is considered visible, and the transaction is set in advance to be visible to the target transaction.

Exemplarily, the global transaction consistency judgment criterion includes: comparing the commit timestamp commit_ts of one transaction with the start timestamp start_ts of another transaction, and performing visibility judgment, if transaction T1.commit_ts<transaction T2.start_ts. Transaction T2 is visible, otherwise not.

Judging by the global consistency, the transaction deleted from the active transaction table indicates that the transaction has been committed, but due to the two-phase commit corresponding to the distributed transaction, the commit phase has not been executed on the local node, so the local transaction is still active table. After the global consistency determines that the uncommitted transaction is set to be visible, the uncommitted transaction status is set to commit completed in advance. When the subsequent submission process of this uncommitted transaction comes, it does not need to be set to be visible again during the transaction submission process. In this embodiment, by setting the visibility in advance, the globally consistent visibility determination is satisfied, and the inefficiency caused by waiting for submission is avoided.

Among them, determine the transactions that need to be added to the local active transaction information, see the following embodiments for details:

In a possible implementation, according to the status information of the transaction and the first start timestamp, the transaction is added to the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction, including:

Determine the submitted transaction whose status information is completed and the commit timestamp is greater than the first start timestamp, if the identification number of the committed transaction is less than the minimum identification number in the local active transaction information, and the old version data of the submitted transaction has been cleared , the execution of the target transaction is terminated, otherwise, the committed transaction is added to the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction.

In practical applications, the status information of transactions in the shared memory is traversed, and the committed transactions whose status information is submitted and whose submission timestamp is greater than the start timestamp of the target transaction are filtered out. Since the identification number of the transaction is allocated incrementally, it has If the identification number of the committed transaction is less than the minimum identification number in the snapshot of the local active transaction list, it indicates that the execution time of the committed transaction is earlier than the transaction in the current local active transaction information. If the old version data of the committed transaction has been cleared, the execution of the target transaction is terminated. If it is added to the snapshot of the active transaction table, according to the MVCC mechanism of the active transaction table, the current version of the data is invisible to the target transaction, but the old version of the data has been cleaned up, which will lead to data errors. Otherwise, the committed transaction is added to the snapshot of the local list of active transactions.

In a possible implementation, the method further includes:

According to the isolation level, determine the modification times of the local active transaction information.

In practical applications, if the target transaction is a distributed transaction, the CN needs to modify the local active transaction information multiple times.

Among them, the isolation level indicates the degree to which a transaction must be isolated from resource or data changes made by other transactions. In this embodiment, the isolation levels may include: Read Commit (Read Commit, RC), Repeatable Read (RepeatableRead, RR) and Serializable Snapshot Isolation (Serializable Snapshot Isolation, SSI). The isolation level can be configured according to specific needs.

Exemplarily, for distributed transactions and non-distributed transactions at the RR or SSI isolation level, you can modify the local active transaction information when executing the first statement of the transaction to obtain globally consistent active transaction information. After executing the transaction statement, directly use the globally consistent active transaction information without regenerating it. For distributed transactions and non-distributed transactions at the RC isolation level, when each statement of the transaction is executed, the local active transaction information must be modified to generate a globally consistent active transaction information.

In a possible implementation manner, the target transaction is a distributed transaction, and the method further includes:

After the target transaction is ready, obtain the submission timestamp from TSO; send a submission operation request to the data node DN in the database, the submission operation request carries the submission timestamp, and execute the submission operation after receiving the submission success notification message sent by DN .

In practical applications, if the target transaction is a distributed transaction, when the statement execution of the transaction needs to be committed, send a submission timestamp acquisition request to TSO, receive the submission timestamp sent by TSO, send a submission operation request to DN, and submit the operation request Carry the submission timestamp in DN, record the submission timestamp and the identification information of the target transaction, execute the submission operation, send a submission success notification message to CN after the submission is completed, CN executes the submission operation, and carries the first start timestamp, and submits the target transaction synchronously state information to TSO, TSO deletes the committed transaction from the cache.

In a possible implementation, the method further includes:

When obtaining the start timestamp or commit timestamp of multiple target transactions from TSO at the same time, combine multiple timestamp acquisition requests into one request and send it to TSO, and receive the same start timestamp or commit time of multiple target transactions sent by TSO stamp.

In practical applications, when a transaction in CN wants to apply for a start time stamp, it is possible that other transactions are also applying for a start time stamp at the same time, or another transaction may be applying for a submission time stamp, or synchronously submitting status information to TSO , these operations can be combined into one request message sent to TSO. When TSO replies to the merge request, it can uniformly send a start timestamp as the start timestamp of multiple transactions applied at the same time, and the commit timestamp is handled in the same way. In this embodiment, by merging applications, the application frequency can be reduced and the pressure on TSO can be reduced. TSO does not need to assign a timestamp to each transaction, which can reduce the pressure on TSO and save network bandwidth.

The embodiment of the present application provides a transaction execution method, as shown in Figure 3 is a flow chart of the transaction execution method in an embodiment of the present application, the method is applied to the data node DN in the database, the method includes:

Step S301, receiving the first start time stamp and the time stamp list of the target transaction sent by the CN or TSO, the time stamp list includes the second start time stamp of the transaction that has completed the preparation but has not completed the commit before the first start time stamp.

Step S302, acquiring local active transaction information; the transactions in the local active transaction information include uncommitted transactions.

Step S303, according to the first start time stamp and the time stamp list, modify the local active transaction information to obtain globally consistent active transaction information corresponding to the target transaction.

Step S304, execute the target transaction according to the globally consistent active transaction information, so as to realize the consistency of the global transaction.

Wherein, if the target transaction is a distributed transaction, the DN receives the first start timestamp and the timestamp list of the target transaction sent by the CN, and the first start timestamp and the timestamp list are acquired by the CN from the TSO. TSO provides global centralized timing services, provides start timestamps for each transaction in a linear increment manner, and provides commit timestamps for distributed transactions. If the target transaction is a non-distributed transaction, the DN receives the first start timestamp and the timestamp list of the target transaction sent by the TSO.

After receiving the first start timestamp and the timestamp list, the DN acquires local active transaction information. The specific form of the local active transaction information is not limited, for example, it may be a snapshot of the local active transaction list. The local active transaction list includes the identification information of the active transaction.

Wherein, the target transaction may be a distributed transaction or a non-distributed transaction. Transaction (Transaction) refers to a series of operations performed as a single logical unit of work, which can include multiple statements, either completely executed or not executed at all. By grouping a set of related operations into an all-or-nothing unit, error recovery is simplified and applications are made more reliable. For a logical unit of work to be a transaction, it needs to satisfy the properties of atomicity, consistency, isolation, and durability. Distributed Transaction means that transaction participants, transaction-supporting devices, resource devices, and transaction managers are located on different nodes of different distributed systems. Distributed transactions can query data across nodes, and non-distributed transactions are transactions that complete data queries within a single node.

The status information of the transaction includes: in process (Inprocess), completed preparation (prepared), committing (committing), and completed committing (committed). The local active transaction information includes the identification information of the uncommitted transaction, and the uncommitted transaction includes the transaction whose state information is executing, completing preparation and committing.

Exemplarily, the DN modifies the snapshot of the local active transaction list according to the start timestamp and the timestamp list sent by the CN, obtains the snapshot of the globally consistent active transaction list corresponding to the target transaction, and executes according to the snapshot of the globally consistent active transaction list Target transaction to achieve data consistency when multiple transactions are executed concurrently in at least one node.

The transaction execution method provided in this embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the time stamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

Wherein, the specific implementation of step S303 is shown in the following embodiments:

In a possible implementation, the local active transaction information is modified according to the first start timestamp and the timestamp list to obtain globally consistent active transaction information corresponding to the target transaction, including:

According to the status information of the transaction, the first start timestamp and the timestamp list, the transaction is deleted and/or added to the local active transaction information, and the globally consistent active transaction information corresponding to the target transaction is obtained.

In practical application, the DN obtains the status information of the transaction from the shared cache interval, and the status information of the transaction includes: Inprocess, Prepared, Committing, Committed. According to the status information of the transaction, the timestamp in the timestamp list and the start timestamp of the target transaction, filter the transaction, delete the identification information of the filtered transaction in the snapshot of the local active transaction list, and/or filter out The identification information of the transaction is added to the snapshot of the local active transaction list, and the identification information of the transaction may be the global identification xid of the transaction. In this embodiment, by deleting and/or adding transactions in the local active transaction information, the globally consistent active transaction information corresponding to the target transaction is obtained, which is used for the execution of the target transaction. By modifying the local active transaction information, the stand-alone database kernel is retained The MVCC based on the active transaction table in the middle does not need the MVCC development based on the timestamp, which reduces the modification of the kernel, and avoids the concurrency conflict caused by the misalignment of the write timestamp and the read timestamp, and realizes high expansion and high compatibility of the database kernel. and high stability.

Among them, the transaction to be deleted is determined in the local active transaction information, see the following embodiments for details:

In a possible implementation, according to the status information of the transaction, the first start timestamp and the timestamp list, the transaction is deleted from the local active transaction information, and the globally consistent active transaction information corresponding to the target transaction is obtained, including:

In the local active transaction information, perform at least one of the following operations to obtain the globally consistent active transaction information corresponding to the target transaction: delete the transaction whose status information is submitting and whose commit timestamp is less than the first start timestamp; the commit timestamp is from Obtained by TSO; delete the transaction whose status information is completed and whose start timestamp is in the timestamp list.

Exemplarily, traversing the status information of the transactions in the shared memory, the transactions deleted in the snapshot of the local active transaction list include at least one of the following: the status information of the transaction is submitting, and the commit timestamp is less than the start timestamp of the target transaction ; The status information of the transaction is ready for completion, and the start timestamp of the transaction is in the timestamp list.

In a possible implementation, the method further includes:

Make transactions deleted in the local active transaction information visible to the target transaction.

In practical applications, pre-visibility settings are performed for transactions deleted from the snapshot of the local active transaction list, because deletion from the snapshot means that the transaction is visible to the target transaction, but because it has not yet completed the commit, it is visible according to the global consensus It is determined that the data is visible, and the transaction is set in advance to be visible to the target transaction.

Judging by the global consistency, the transaction deleted from the active transaction table indicates that the transaction has been committed, but due to the two-phase commit corresponding to the distributed transaction, the commit phase has not been executed on the local node, so the local transaction is still active table. After the global consistency determines that the uncommitted transaction is set to be visible, the uncommitted transaction status is set to commit completed in advance. When the subsequent submission process of this uncommitted transaction comes, it does not need to be set to be visible again during the transaction submission process. In this embodiment, by setting the visibility in advance, the globally consistent visibility determination is satisfied, and the inefficiency caused by waiting for submission is avoided.

Among them, determine the transactions that need to be added to the local active transaction information, see the following embodiments for details:

In a possible implementation, according to the status information of the transaction and the first start timestamp, the transaction is added to the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction, including:

Determine the submitted transaction whose status information is completed and the commit timestamp is greater than the first start timestamp, if the identification number of the committed transaction is less than the minimum identification number in the local active transaction information, and the old version data of the submitted transaction has been cleared , the execution of the target transaction is terminated, otherwise, the committed transaction is added to the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction.

In practical applications, the status information of transactions in the shared memory is traversed, and the committed transactions whose status information is submitted and whose submission timestamp is greater than the start timestamp of the target transaction are filtered out. Since the identification number of the transaction is allocated incrementally, it has If the identification number of the committed transaction is less than the minimum identification number in the snapshot of the local active transaction list, it indicates that the execution time of the committed transaction is earlier than the transaction in the current local active transaction information. If the old version data of the committed transaction has been cleared, the execution of the target transaction is terminated. If it is added to the snapshot of the active transaction table, according to the MVCC mechanism of the active transaction table, the current version data is invisible to the target transaction, but the old version data has been cleaned up, which will cause data errors. Otherwise, the committed transaction is added to the snapshot of the local list of active transactions.

In a possible implementation, the method further includes:

According to the isolation level, determine the modification times of the local active transaction information.

In practical applications, if the target transaction is a distributed transaction, the DN needs to modify the local active transaction information multiple times.

Among them, the isolation level indicates the degree to which a transaction must be isolated from resource or data changes made by other transactions. In this embodiment, the isolation levels may include: Read Commit (Read Commit, RC), Repeatable Read (RepeatableRead, RR) and Serializable Snapshot Isolation (Serializable Snapshot Isolation, SSI). The isolation level can be configured according to specific needs.

Exemplarily, for distributed transactions and non-distributed transactions at the RR or SSI isolation level, you can modify the local active transaction information when executing the first statement of the transaction to obtain globally consistent active transaction information. After executing the transaction statement, directly use the globally consistent active transaction information without regenerating it. For distributed transactions and non-distributed transactions at the RC isolation level, when each statement of the transaction is executed, the local active transaction information must be modified to generate a globally consistent active transaction information.

In a possible implementation manner, the target transaction is a distributed transaction, and the method further includes:

Receive the commit operation request sent by CN, carry the commit timestamp of the target transaction in the commit operation request, record the commit timestamp, execute the commit operation, and send a commit success notification message to CN, so that CN can execute the commit operation.

In practical applications, if the target transaction is a distributed transaction, when the statement execution of the transaction needs to be submitted, CN sends a submission timestamp acquisition request to TSO, receives the submission timestamp sent by TSO, sends a submission operation request to DN, and submits the operation The request carries the submission timestamp, and the DN receives the submission operation request sent by the CN, records the submission timestamp and the identification information of the target transaction, executes the submission operation, and sends a submission success notification message to the CN after the submission is completed, and the CN executes the submission operation and carries the first At the beginning of the timestamp, the status information of the target transaction is synchronously submitted to the TSO, and the TSO deletes the committed transaction from the cache.

In a possible implementation, the method further includes:

When obtaining the start timestamp or commit timestamp of multiple target transactions from TSO at the same time, combine multiple timestamp acquisition requests into one request and send it to TSO, and receive the same start timestamp or commit time of multiple target transactions sent by TSO stamp.

In practical applications, when a transaction in the DN wants to apply for a start timestamp, it is possible that other transactions are also applying for a start timestamp at the same time, or another transaction may apply for a commit timestamp, or submit status information to TSO synchronously , these operations can be combined into one request message sent to TSO. When TSO replies to the merge request, it can uniformly send a start timestamp as the start timestamp of multiple transactions applied at the same time, and the commit timestamp is handled in the same way. In this embodiment, by merging applications, the application frequency can be reduced and the pressure on TSO can be reduced. TSO does not need to assign a timestamp to each transaction, which can reduce the pressure on TSO and save network bandwidth.

The embodiment of the present application provides a transaction execution method, as shown in Figure 4 is a flow chart of the transaction execution method in an embodiment of the present application, the method is applied to the timing node TSO in the database, the method includes:

Step S401, receiving the start timestamp acquisition request of the target transaction sent by CN or DN.

Step S402, send the first start timestamp and timestamp list of the target transaction to CN or DN, so that CN or DN can modify the local active transaction information according to the first start timestamp and timestamp list, and obtain the global consistency corresponding to the target transaction Sexually active transaction information; the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp; the globally consistent active transaction information is used for CN or DN to execute the target transaction to achieve global transactional consistency.

Wherein, if the target transaction is a distributed transaction, the TSO receives the start timestamp acquisition request of the target transaction sent by the CN, and sends the first start timestamp and the timestamp list of the target transaction to the CN. If the target transaction is a non-distributed transaction, the TSO receives the start timestamp acquisition request of the target transaction sent by the CN or DN, and sends the first start timestamp and timestamp list of the target transaction to the CN or DN.

The transaction execution method provided in this embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the time stamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

In a possible implementation manner, the target transaction is a distributed transaction, and the method further includes:

Receive the submission timestamp acquisition request of the target transaction sent by CN, generate the submission timestamp of the target transaction and send it to CN.

In practical applications, if the target transaction is a distributed transaction, when the statement execution of the transaction needs to be submitted, CN sends a submission timestamp acquisition request to TSO, TSO generates the submission timestamp of the target transaction and sends it to CN, and CN executes the submission operation. And carry the first start time stamp, synchronously submit the state information of the target transaction to TSO, and TSO deletes the committed transaction from the cache.

In an example, the meanings and service functions of each field of TSO are as follows:

i.Data structure:

1.current_ts//used to record the current timestamp.

2.tso_prepared//Used to cache the start_ts for completing the registration transaction.

ii. Centralized time service:

1. TSO uniformly assigns the SQL statement start timestamp start_ts=current_ts++ for each SQL statement in a transaction in a linearly increasing manner.

2. TSO uniformly assigns a commit timestamp commit_ts=current_ts++ to each prepared distributed transaction in a linearly increasing manner.

iii. Completion of the prepared transaction registration management service:

1. TSO provides registration services for each prepared distributed transaction, that is, at the same time as assigning the commit timestamp commit_ts, the transaction start timestamp start_ts, which is the timestamp obtained by the first statement after the transaction is opened, is recorded in tso_prepared middle.

2. TSO provides a logout service for each distributed transaction that has been submitted, that is, the start_ts recorded when applying for commit_ts before logout, that is, clearing the corresponding record of tso_prepared.

TSO provides a prepared transaction notification service for each transaction that applies for start_ts, that is, returns all start_ts cached in tso_prepared as a timestamp list start_ts_plist to the transaction that applies for start_ts.

In an example, the meanings of the fields involved in the process of modifying the snapshot of the local active transaction list by CN or DN are as follows:

data structure:

1.tso_status//It is used to record the status information of the transaction in TSO and the snapshot of the global consistency list after the transaction is started in each node.

2.status.//Used to record the current transaction status information. Including Inprocess, prepared, committing, and committed.

3.current_start_ts//The start timestamp of the current transaction used to record the transaction.

4.commit_ts//Used to record the commit timestamp of the current transaction.

5.snapshot//Used to record the snapshot of the local global consistency active transaction table that conforms to the start timestamp start_ts and the timestamp list start_ts_plist.

6.tso_commit//Used to record the transaction information after the commit is completed, including the commit_ts of the transaction, and the global identifier xid information of the corresponding transaction.

In a distributed transaction, each Structured Query Language (SQL) statement adopts a unified start_ts and start_ts_plist at each of its participating nodes. That is, the start timestamp of the SQL statement and the completed preparation registered in the TSO at the current moment are not yet submitted. A list of start timestamps for transactions.

Corresponding to the above method embodiments, the embodiment of the present application also provides a transaction execution method, the method includes:

Step S11, the timing node TSO receives the start timestamp acquisition request of the target transaction sent by the coordinating node CN or the data node DN.

In step S12, the TSO sends the first start time stamp and the time stamp list of the target transaction to the CN or DN; the time stamp list includes the second start time stamp of the transaction that has completed the preparation but has not completed the commit before the first start time stamp.

In step S13, the CN or DN modifies the local active transaction information according to the first start timestamp and the timestamp list, and obtains globally consistent active transaction information corresponding to the target transaction.

In step S14, the CN or DN executes the target transaction according to the globally consistent active transaction information, so as to achieve global transaction consistency.

Wherein, if the target transaction is a distributed transaction, the TSO receives the start timestamp acquisition request of the target transaction sent by the CN, and sends the first start timestamp and the timestamp list of the target transaction to the CN. If the target transaction is a non-distributed transaction, the TSO receives the start timestamp acquisition request of the target transaction sent by the CN or DN, and sends the first start timestamp and timestamp list of the target transaction to the CN or DN.

The coordination node CN receives the execution request of the target transaction and obtains the local active transaction information; the transaction in the local active transaction information includes the uncommitted transaction; obtains the first start timestamp and timestamp list of the target transaction from the timing node TSO in the database , the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp; according to the first start timestamp and the timestamp list, modify the local active transaction information to obtain the target transaction corresponding Global consistency active transaction information; according to the global consistency active transaction information, execute the target transaction to achieve global transaction consistency.

The data node DN receives the first start timestamp and timestamp list of the target transaction sent by CN or TSO, and the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp; obtain Local active transaction information; the transactions in the local active transaction information include uncompleted committed transactions; according to the first start timestamp and timestamp list, modify the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction; according to the global Consistency active transaction information, execute the target transaction to achieve global transaction consistency.

Wherein, the specific working processes of the timing node TSO, the coordinating node CN and the data node DN are the same as those of the above-mentioned method embodiments, and will not be repeated here.

The transaction execution method provided in this embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the time stamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

In order to introduce the technical idea of the technical solution of this application more clearly, the following will introduce the execution process of distributed transactions and non-distributed transactions according to different isolation levels:

FIG. 5 is an interaction diagram of a distributed transaction execution process at the RR/SSI isolation level provided by the embodiment of the present application. As shown in Figure 5, the execution process includes:

Step 501, the coordinating node CN receives the execution request of the target transaction, and starts to execute the first SQL statement.

Step 502, the coordinating node CN obtains a snapshot of the local active transaction list.

Step 503, sending a request for acquiring the start timestamp of the target transaction to the timing node.

Step 504, generate the start timestamp of the target transaction, and obtain the timestamp list from the cache.

Step 505, sending the start time stamp and time stamp list of the target transaction.

Step 506, setting the state of the target transaction as executing. According to the start timestamp of the target transaction, the timestamp list and the snapshot of the local active transaction list, a global snapshot is generated, that is, globally consistent active transaction information.

Step 507, sending the start time stamp and time stamp list of the target transaction.

Step 508, setting the status of the target transaction as executing. Generate a global snapshot based on the start timestamp of the target transaction, the timestamp list and the snapshot of the local active transaction list.

Step 509, sending the first SQL statement.

Step 510, using the global snapshot as the snapshot of the target transaction to execute the second SQL statement.

Step 511, sending the second SQL statement.

Step 512, using the global snapshot as the snapshot of the target transaction to execute the first SQL statement.

Step 513, using the global snapshot as the snapshot of the target transaction to execute the second SQL statement.

Step 514, sending a prepare to submit request message.

Step 515, execute the preparation process of the target transaction, and set the state information of the target transaction as preparation completed after the execution is completed.

Step 516, sending a preparation success notification message.

Step 517, set the status information of the target transaction as ready.

Step 518, carry the start timestamp of the target transaction, and apply for the commit timestamp.

Step 519, record the start timestamp of the target transaction, and calculate the commit timestamp.

Step 520, return the commit timestamp of the target transaction.

Step 521, set the state of the target transaction as committing, and record the commit timestamp and identification information.

Step 522, carry the submission time stamp, and apply for the submission operation.

Step 523, setting the status information of the target transaction as submitting. Record the commit timestamp and identification information of the target transaction, and execute the commit operation to delete the status information of the target transaction.

Step 524, sending a submission success notification message.

Step 525, perform a submit operation.

Step 526, carrying the start time stamp of the target transaction, submitting the state information of the target transaction.

Step 527, delete the committed transaction in the cache.

Step 528, delete the status information of the target transaction in the cache.

FIG. 6 is an interactive diagram of the distributed transaction execution process at the RC isolation level provided by the embodiment of the present application. As shown in Figure 6, the execution process includes:

Step 601, receiving an execution request of a target transaction, and starting to execute a first SQL statement.

Step 602, acquire a snapshot of the local active transaction list.

Step 603, sending a request for acquiring the start timestamp of the target transaction to the timing node.

Step 604, generate the start timestamp of the target transaction, and obtain the timestamp list from the cache.

Step 605, sending the start time stamp and time stamp list of the target transaction.

Step 606, setting the status of the target transaction as executing. According to the start timestamp of the target transaction, the timestamp list and the snapshot of the local active transaction list, a global snapshot is generated, that is, globally consistent active transaction information.

Step 607, sending the start time stamp and time stamp list of the target transaction.

Step 608, setting the status of the target transaction as executing. Generate a global snapshot based on the start timestamp of the target transaction, the timestamp list and the snapshot of the local active transaction list.

Step 609, sending the first SQL statement.

Step 610, using the global snapshot as the snapshot of the target transaction to execute the first SQL statement.

Step 611, execute the second SQL statement to obtain a snapshot of the local active transaction list.

Step 612, sending a request for obtaining the start timestamp of the target transaction to the timing node.

Step 613, generate the start timestamp of the target transaction, and obtain the timestamp list from the cache.

Step 614, sending the start time stamp and time stamp list of the target transaction.

Step 615: Generate a global snapshot according to the target transaction's start time stamp, the time stamp list and the snapshot of the local active transaction list.

Step 616, sending the start time stamp and time stamp list of the target transaction.

Step 617, generate a global snapshot according to the snapshot of the target transaction start timestamp, timestamp list, and local active transaction list.

Step 618, sending the second SQL statement.

Step 619, using the global snapshot as the snapshot of the target transaction to execute the second SQL statement.

Step 620, sending a ready-to-submit notification message.

Step 621, execute the transaction preparation process, and set the state information of the target transaction as preparation completed.

Step 622, sending a submission success notification message.

Step 623, set the status information of the target transaction as ready.

Step 624, carry the start timestamp of the target transaction, and apply for the commit timestamp.

Step 625, record the start timestamp of the target transaction, and calculate the commit timestamp.

Step 626, return the commit timestamp of the target transaction.

Step 627, set the state information of the target transaction as submitting, and record the commit timestamp and identification information of the target transaction.

Step 628, carry the submission time stamp, and apply for the submission operation.

Step 629, setting the status information of the target transaction as submitting. Record the commit timestamp and identification information of the target transaction, and execute the commit operation to delete the status information of the target transaction.

Step 630, sending a submission success notification message.

Step 631, perform a submit operation.

Step 632, carry the start timestamp of the target transaction, and commit the state information synchronously.

Step 633, delete the committed transaction in the cache.

Step 634, delete the state information of the target transaction in the cache.

FIG. 7 is an interaction diagram of a non-distributed transaction execution process at the RR/SSI isolation level provided by the embodiment of the present application. As shown in Figure 7, the execution process includes:

Step 701, receiving an execution request of a target transaction, and starting to execute a first SQL statement.

Step 702, acquire a snapshot of the local active transaction list.

Step 703, sending a request for obtaining the start timestamp of the target transaction to the timing node.

Step 704, generate the start timestamp of the target transaction, and obtain the timestamp list from the cache.

Step 705, sending the start time stamp and time stamp list of the target transaction.

Step 706, setting the state of the target transaction as executing. According to the start timestamp of the target transaction, the timestamp list and the snapshot of the local active transaction list, a global snapshot is generated, that is, globally consistent active transaction information.

Step 707, using the global snapshot as the snapshot of the target transaction to execute the second SQL statement.

Step 708, execute the commit operation of the target transaction, and delete the state information of the target transaction.

FIG. 8 is an interactive diagram of the non-distributed transaction execution process at the RC isolation level provided by the embodiment of the present application. As shown in Figure 8, the execution process includes:

Step 801, receiving an execution request of a target transaction, and starting to execute a first SQL statement.

Step 802, acquire a snapshot of the local active transaction list.

Step 803, sending a request for obtaining the start timestamp of the target transaction to the timing node.

Step 804, generate the start timestamp of the target transaction, and obtain the timestamp list from the cache.

Step 805, sending the start time stamp and time stamp list of the target transaction.

Step 806, setting the status of the target transaction as being executed. According to the start timestamp of the target transaction, the timestamp list and the snapshot of the local active transaction list, a global snapshot is generated, that is, globally consistent active transaction information.

Step 807, execute the second SQL statement to obtain a snapshot of the local active transaction list.

Step 808, sending a request for acquiring the start timestamp of the target transaction to the timing node.

Step 809, generate the start timestamp of the target transaction, and obtain the timestamp list from the cache.

Step 810, sending the start time stamp and time stamp list of the target transaction.

Step 811 , generate a global snapshot according to the start timestamp of the target transaction, the timestamp list and the snapshot of the local active transaction list.

Step 812, execute the commit operation of the target transaction, and delete the state information of the target transaction.

Corresponding to the method applied to the CN provided in the embodiment of the present application, the embodiment of the present application further provides a transaction execution apparatus. FIG. 9 is a structural block diagram of a transaction execution device according to an embodiment of the present application, which may include:

The receiving module 901 is configured to receive the execution request of the target transaction, and acquire local active transaction information; the transactions in the local active transaction information include uncompleted committed transactions;

The acquiring module 902 is configured to acquire the first start timestamp and the timestamp list of the target transaction from the timing node TSO in the database, and the timestamp list includes the second transaction of the transaction that has completed the preparation and has not yet completed the commit before the first start timestamp. start timestamp;

The modification module 903 is configured to modify the local active transaction information according to the first start time stamp and the time stamp list, and obtain globally consistent active transaction information corresponding to the target transaction;

The execution module 904 is configured to execute the target transaction according to the globally consistent active transaction information, so as to realize the consistency of the global transaction.

The transaction execution device provided in this embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the timestamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

In a possible implementation, module 903 is modified to:

According to the status information of the transaction, the first start timestamp and the timestamp list, the transaction is deleted and/or added to the local active transaction information, and the globally consistent active transaction information corresponding to the target transaction is obtained.

In a possible implementation, the modification module 903 deletes the transaction from the local active transaction information according to the state information of the transaction, the first start timestamp and the timestamp list, and obtains the globally consistent active transaction information corresponding to the target transaction , for:

In the local active transaction information, perform at least one of the following operations to obtain the globally consistent active transaction information corresponding to the target transaction: delete the transaction whose status information is submitting and whose commit timestamp is less than the first start timestamp; the commit timestamp is from Obtained by TSO; delete the transaction whose status information is completed and whose start timestamp is in the timestamp list.

In a possible implementation manner, the device further includes a setting module, configured to:

Make transactions deleted in the local active transaction information visible to the target transaction.

In a possible implementation, the modification module 903 is used to:

Determine the submitted transaction whose status information is completed and the commit timestamp is greater than the first start timestamp, if the identification number of the committed transaction is less than the minimum identification number in the local active transaction information, and the old version data of the submitted transaction has been cleared , the execution of the target transaction is terminated, otherwise, the committed transaction is added to the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction.

In a possible implementation, the modifying module 903 is also used to:

According to the isolation level, determine the modification times of the local active transaction information.

In a possible implementation, the target transaction is a distributed transaction, and the acquisition module 902 is also used to:

After the target transaction is ready, obtain the submission timestamp from TSO; send a submission operation request to the data node DN in the database, the submission operation request carries the submission timestamp, and execute the submission operation after receiving the submission success notification message sent by DN .

In a possible implementation, the device further includes a merging module, configured to:

When obtaining the start timestamp or commit timestamp of multiple target transactions from TSO at the same time, combine multiple timestamp acquisition requests into one request and send it to TSO, and receive the same start timestamp or commit time of multiple target transactions sent by TSO stamp.

The functions of each module in each device in the embodiment of the present application can refer to the corresponding description in the above method, and have corresponding beneficial effects, and will not be repeated here.

Corresponding to the method applied to the DN provided by the embodiment of the present application, the embodiment of the present application further provides a transaction execution apparatus. FIG. 10 is a structural block diagram of a transaction execution device according to an embodiment of the present application, which may include:

The receiving module 1001 is configured to receive the first start time stamp and the time stamp list of the target transaction sent by CN or TSO, and the time stamp list includes the second start time of the transaction that has completed the preparation and has not completed the commit before the first start time stamp stamp;

An acquisition module 1002, configured to acquire local active transaction information; the transactions in the local active transaction information include unfinished committed transactions;

The modifying module 1003 is configured to modify the local active transaction information according to the first start timestamp and the timestamp list, and obtain globally consistent active transaction information corresponding to the target transaction;

The execution module 1004 is configured to execute the target transaction according to the global consistency active transaction information, so as to realize the consistency of the global transaction.

The transaction execution device provided in this embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the timestamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

In a possible implementation, module 1003 is modified to:

According to the status information of the transaction, the first start timestamp and the timestamp list, the transaction is deleted and/or added to the local active transaction information, and the globally consistent active transaction information corresponding to the target transaction is obtained.

In a possible implementation, the modification module 1003 deletes the transaction from the local active transaction information according to the state information of the transaction, the first start timestamp and the timestamp list, and obtains the globally consistent active transaction information corresponding to the target transaction , for:

In the local active transaction information, perform at least one of the following operations to obtain the globally consistent active transaction information corresponding to the target transaction: delete the transaction whose status information is submitting and whose commit timestamp is less than the first start timestamp; the commit timestamp is from Obtained by TSO; delete the transaction whose status information is completed and whose start timestamp is in the timestamp list.

In a possible implementation manner, the device further includes a setting module, configured to:

Make transactions deleted in the local active transaction information visible to the target transaction.

In a possible implementation, the modification module 1003 is used to:

Determine the submitted transaction whose status information is completed and the commit timestamp is greater than the first start timestamp, if the identification number of the committed transaction is less than the minimum identification number in the local active transaction information, and the old version data of the submitted transaction has been cleared , the execution of the target transaction is terminated, otherwise, the committed transaction is added to the local active transaction information to obtain the globally consistent active transaction information corresponding to the target transaction.

In a possible implementation, the modifying module 1003 is also used to:

According to the isolation level, determine the modification times of the local active transaction information.

In a possible implementation, the target transaction is a distributed transaction, and the receiving module 1001 is also used to:

Receive the commit operation request sent by CN, carry the commit timestamp of the target transaction in the commit operation request, record the commit timestamp, execute the commit operation, and send a commit success notification message to CN, so that CN can execute the commit operation.

The functions of each module in each device in the embodiment of the present application can refer to the corresponding description in the above method, and have corresponding beneficial effects, and will not be repeated here.

Corresponding to the method applied to the TSO provided by the embodiment of the present application, the embodiment of the present application further provides a transaction execution device. FIG. 11 is a structural block diagram of a transaction execution device according to an embodiment of the present application, which may include:

The receiving module 1101 is configured to receive the start timestamp acquisition request of the target transaction sent by the CN or DN.

The sending module 1102 is configured to send the first start timestamp and the timestamp list of the target transaction to the CN or DN, so that the CN or DN modifies the local active transaction information according to the first start timestamp and the timestamp list, and obtains the target transaction corresponding The globally consistent active transaction information; the timestamp list includes the second start timestamp of the transaction that has completed the preparation and has not completed the commit before the first start timestamp; the globally consistent active transaction information is used for CN or DN to execute the target transaction, To achieve global transaction consistency.

The transaction execution device provided in this embodiment integrates the TSO and local active transaction information, and modifies the local active transaction information according to the timestamp assigned by the TSO to achieve global transaction consistency. Compared with related technologies, through the TSO global centralized timing service, the length of the time stamp is fixed, and will not become larger due to the increase of transactions, and will not affect the scalability of the distributed cluster. Moreover, by modifying the local active transaction information, the MVCC based on the active transaction table in the kernel of the stand-alone database is retained, and there is no need for timestamp-based MVCC development, which reduces the modification of the kernel and avoids the misalignment of the write timestamp and the read timestamp. The concurrency conflict achieves high expansion, high compatibility and high stability of the database kernel.

In a possible implementation, the target transaction is a distributed transaction, and the device further includes a generating module, configured to:

Receive the submission timestamp acquisition request of the target transaction sent by CN, generate the submission timestamp of the target transaction and send it to CN.

The functions of each module in each device in the embodiment of the present application can refer to the corresponding description in the above method, and have corresponding beneficial effects, and will not be repeated here.

The embodiment of the present application also provides a distributed database system, and the system includes the coordination node CN, the data node DN and the timing node TSO in any of the above embodiments. Wherein, there may be at least one coordinating node CN and data node DN.

For the functions of each node in the system provided by the embodiment of the present application, refer to the corresponding description in the above method, and have corresponding beneficial effects, which will not be repeated here.

FIG. 12 is a block diagram of an electronic device used to implement an embodiment of the present application. As shown in FIG. 12 , the electronic device includes: a memory 1210 and a processor 1220 , and the memory 1210 stores computer programs that can run on the processor 1220 . When the processor 1220 executes the computer program, the methods in the foregoing embodiments are implemented. The number of memory 1210 and processor 1220 may be one or more.

This electronic device also includes:

The communication interface 1230 is used to communicate with external devices for interactive data transmission.

If the memory 1210, the processor 1220, and the communication interface 1230 are implemented independently, the memory 1210, the processor 1220, and the communication interface 1230 may be connected to each other through a bus to complete mutual communication. The bus may be an Industry Standard Architecture (Industry Standard Architecture, ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (Extended Industry Standard Architecture, EISA) bus, etc. The bus can be divided into address bus, data bus, control bus and so on. For ease of representation, only one thick line is used in FIG. 12 , but it does not mean that there is only one bus or one type of bus.

Optionally, in a specific implementation, if the memory 1210, the processor 1220, and the communication interface 1230 are integrated on one chip, the memory 1210, the processor 1220, and the communication interface 1230 may communicate with each other through the internal interface.

The embodiment of the present application provides a computer-readable storage medium, which stores a computer program, and implements the method provided in the embodiment of the present application when the program is executed by a processor.

The embodiment of the present application also provides a chip, the chip includes a processor, configured to call and execute instructions stored in the memory from the memory, so that the communication device installed with the chip executes the method provided in the embodiment of the present application.

The embodiment of the present application also provides a chip, including: an input interface, an output interface, a processor, and a memory, the input interface, the output interface, the processor, and the memory are connected through an internal connection path, and the processor is used to execute the code in the memory , when the code is executed, the processor is used to execute the method provided by the embodiment of the application.

It should be understood that the above-mentioned processor may be a central processing unit (Central Processing Unit, CPU), and may also be other general-purpose processors, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), Field Programmable Gate Array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general purpose processor may be a microprocessor or any conventional processor or the like. It should be noted that the processor may be a processor supporting Advanced RISC Machines (ARM) architecture.

Further, optionally, the foregoing memory may include a read-only memory and a random access memory. The memory can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. Among them, the non-volatile memory can include read-only memory (Read-Only Memory, ROM), programmable read-only memory (Programmable ROM, PROM), erasable programmable read-only memory (Erasable PROM, EPROM), electrically programmable Erase Programmable Read-Only Memory (Electrically EPROM, EEPROM) or Flash. Volatile memory can include Random Access Memory (RAM), which acts as external cache memory. By way of illustration and not limitation, many forms of RAM are available. For example, static random access memory (Static RAM, SRAM), dynamic random access memory (Dynamic Random Access Memory, DRAM), synchronous dynamic random access memory (Synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (Double Data RateSDRAM, DDR SDRAM), Enhanced Synchronous Dynamic Random Access Memory (Enhanced SDRAM, ESDRAM), Synchronous Link Dynamic Random Access Memory (Sync link DRAM, SLDRAM) and Direct Memory Bus Random Access Memory (DirectRambus RAM, DR RAM).

In the above embodiments, all or part of them may be implemented by software, hardware, firmware or any combination thereof. When implemented using software, it may be implemented in whole or in part in the form of a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the processes or functions according to the present application are produced in whole or in part. A computer can be a general purpose computer, special purpose computer, computer network, or other programmable device. Computer instructions may be stored in, or transmitted from, one computer-readable storage medium to another computer-readable storage medium.

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present application. Furthermore, the described specific features, structures, materials or characteristics may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples and features of different embodiments or examples described in this specification without conflicting with each other.

In addition, the terms "first" and "second" are used for descriptive purposes only, and cannot be interpreted as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of these features. In the description of the present application, "plurality" means two or more, unless otherwise specifically defined.

Any process or method described in a flowchart or otherwise described herein may be understood as representing a module, segment, or code comprising one or more executable instructions for implementing a specific logical function or step of the process part. Also, the scope of preferred embodiments of the present application includes additional implementations in which functions may be performed out of the order shown or discussed, including substantially concurrently or in reverse order as the functions involved are involved.

The logic and/or steps described in the flowcharts or otherwise described herein, for example, can be considered as a sequenced listing of executable instructions for implementing logical functions, which can be embodied in any computer-readable medium , for use with an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, apparatus, or device and execute instructions), or in conjunction with such an instruction execution system, device or equipment.

It should be understood that each part of the present application may be realized by hardware, software, firmware or a combination thereof. In the embodiments described above, various steps or methods may be implemented by software or firmware stored in memory and executed by a suitable instruction execution system. All or part of the steps of the method in the above embodiments can be completed by instructing related hardware through a program. The program can be stored in a computer-readable storage medium. When the program is executed, it includes one of the steps of the method embodiment or its combination.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing module, each unit may exist separately physically, or two or more units may be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software function modules. If the above-mentioned integrated modules are implemented in the form of software function modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. The storage medium may be a read-only memory, a magnetic disk or an optical disk, and the like.

The above is only an exemplary embodiment of the application, but the scope of protection of the application is not limited thereto, and any skilled person familiar with the technical field can easily think of its various changes within the technical scope of the application Or replacement, all of these should be covered within the scope of protection of this application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims (15)

1. A transaction execution method, applied to a coordinating node CN in a database, the method comprising:

receiving an execution request of a target transaction, and acquiring local active transaction information; the transactions in the local active transaction information comprise incomplete committed transactions;

acquiring a first starting timestamp and a timestamp list of the target transaction from a time service node (TSO) in the database, wherein the timestamp list comprises a second starting timestamp of a transaction which is completely prepared and is not completely submitted before the first starting timestamp;

modifying the local active transaction information according to the first starting timestamp and the timestamp list to obtain global consistency active transaction information corresponding to the target transaction;

and executing the target transaction according to the global consistency active transaction information so as to realize the consistency of the global transaction.

2. The method of claim 1, wherein the modifying the locally active transaction information according to the first start timestamp and the timestamp list to obtain global consistent active transaction information corresponding to the target transaction comprises:

and deleting and/or adding the transaction in the local active transaction information according to the state information of the transaction, the first starting timestamp and the timestamp list to obtain global consistent active transaction information corresponding to the target transaction.

3. The method according to claim 2, wherein deleting a transaction in the local active transaction information according to the status information of the transaction, the first start timestamp, and the timestamp list to obtain global consistent active transaction information corresponding to the target transaction, includes:

in the local active transaction information, at least one of the following operations is performed to obtain global consistent active transaction information corresponding to the target transaction:

deleting the transaction of which the state information is in commit and the commit timestamp is smaller than the first start timestamp; the commit timestamp is obtained from the TSO;

deleting the transaction whose state information is completely prepared and whose starting timestamp is in the timestamp list.

4. The method of claim 2, wherein adding a transaction to the locally active transaction information according to the state information of the transaction and the first start timestamp to obtain global consistent active transaction information corresponding to the target transaction comprises:

and determining that the state information is a committed transaction of which the commit is completed and the commit timestamp is greater than the first start timestamp, if the identification number of the committed transaction is smaller than the minimum identification number in the local active transaction information and the old version data of the committed transaction is already cleared, terminating the execution of the target transaction, and if not, adding the committed transaction to the local active transaction information to obtain the global consistency active transaction information.

5. A method according to claim 2 or 3, characterized in that the method further comprises:

setting the deleted transaction in the locally active transaction information to be visible to the target transaction.

6. The method according to any one of claims 1-4, further comprising:

and determining the number of times of modifying the local active transaction information according to the isolation level.

7. The method of any of claims 1-4, wherein the target transaction is a distributed transaction, the method further comprising:

obtaining a commit timestamp from the TSO after the target transaction is ready;

and sending a submission operation request to a data node DN in the database, wherein the submission operation request carries the submission timestamp, and after receiving a submission success notification message sent by the DN, executing the submission operation.

8. The method of claim 7, further comprising:

when the starting time stamps or the submitting time stamps of a plurality of target transactions are obtained from the TSO at the same time, a plurality of time stamp obtaining requests are combined into one request to be sent to the TSO, and the same starting time stamps or the submitting time stamps of the plurality of target transactions sent by the TSO are received.

9. A transaction execution method applied to a data node DN in a database, the method comprising:

receiving a first start timestamp and a timestamp list of a target transaction sent by a coordinating node CN or a time service node TSO, wherein the timestamp list comprises a second start timestamp of a transaction which is prepared and submitted incompletely before the first start timestamp;

acquiring local active transaction information; the transactions in the local active transaction information comprise incomplete committed transactions;

modifying the local active transaction according to the first starting timestamp and the timestamp list to obtain global consistency active transaction information corresponding to the target transaction;

and executing the target transaction according to the global consistency active transaction information so as to realize the consistency of the global transaction.

10. The method of claim 8 or 9, wherein the target transaction is a distributed transaction, the method further comprising:

receiving a commit operation request sent by the CN, wherein the commit operation request carries a commit timestamp of the target transaction, recording the commit timestamp, executing the commit operation, and sending a commit success notification message to the CN so that the CN executes the commit operation.

11. A transaction execution method applied to a time service node (TSO) in a database, the method comprising:

receiving a starting timestamp acquiring request of a target transaction sent by a coordinating node CN or a data node DN;

sending a first starting timestamp and a timestamp list of the target transaction to the CN or the DN, so that the CN or the DN modifies local active transaction information according to the first starting timestamp and the timestamp list to obtain global consistency active transaction information corresponding to the target transaction; the list of timestamps includes a second begin timestamp of a transaction that has completed preparation and not completed commit before the first begin timestamp; the global consistency active transaction information is used for the CN or the DN to execute the target transaction so as to realize the consistency of the global transaction.

12. A method of transaction execution, the method comprising:

a time service node TSO receives a starting timestamp acquisition request of a target transaction sent by a coordinating node CN or a data node DN;

the TSO sending a first start timestamp and a list of timestamps for the target transaction to the CN or the DN; the list of timestamps includes a second start timestamp of a transaction that has completed preparation and has not completed commit before the first start timestamp;

the CN or the DN modifies local active transaction information according to the first starting timestamp and the timestamp list to obtain global consistency active transaction information corresponding to the target transaction;

and the CN or the DN executes the target transaction according to the global consistency active transaction information so as to realize the consistency of the global transaction.

13. A distributed database system, characterized in that the system comprises a coordinating node CN according to any of the preceding claims 1-8, a data node DN according to any of the preceding claims 9-10 and a time service node TSO according to claim 11.

14. An electronic device comprising a memory, a processor and a computer program stored on the memory, the processor implementing the method of any of claims 1-12 when executing the computer program.

15. A computer-readable storage medium, having stored therein a computer program which, when executed by a processor, implements the method of any of claims 1-12.

Images