CN1350673A - Method for transmitting, inside a distributed system, messages between a client instance assigned to a first process and at least one server instance assigned to at least one additional process - Google Patents

Abstract

After receiving a message sent by the client instance (Client) to at least one server instance (Server), a first instance (Object Handler 1) which contains a first process and which is of partner instances provided as mutual communications partners selects at least one appropriate additional instance (Object Handler 2), said additional instance containing the at least one additional process, of the partner instances in order to accept and relay messages. The at least one additional instance containing the at least one additional process relays this message to at least one server instance addressed by the same and optionally receives, from the at least one server instance, a message for relaying to the client instance via the first instance that contains the first process.

Description

In a distributed system between client instances assigned to the first process and assigned to at least Method for transferring information between at least one server instance of another process

The invention relates to a method of transferring information between a client instance assigned to a first process and at least one server instance assigned to at least one other process in a distributed system.

Distributed systems have particular significance in today's telecommunications systems, which are often multiprocessor systems. Distributed systems are characterized in particular by the fact that processes can always be assigned to different processors, which processors can, if necessary, be located on partially separate platforms in the distributed system. Here, for communication between different processes in a distributed system, the most important point is platform transparency. That is, a process intending to send information to another process does not have to know the platform on which the other process is running. Such complex distributed systems must meet many other requirements today. Furthermore, it must be shown to be very reliable, as flexible and open as possible for adaptation and extension. Therefore, the software of such a complex distributed system should be highly modularized and designed outwards with fixed-defined open interfaces, so that each module of the software can be easily matched and reused.

In order to be able to suitably meet the reusability requirements of the above-mentioned software in particular, the software of such a distributed system can be compiled by means of object-oriented design methods and object-oriented programming. But in a distributed system, most of the objects are assigned to different and sometimes simultaneous running processes, and the assignment of these objects to each other cannot be satisfactorily resolved. Sometimes, even a purely object-oriented system design must be interrupted in conventional procedural programming techniques, so that the effect of reusing program parts achieved with object-orientation is more or less lost.

Now discuss the following known measures for introducing simultaneous execution and parallel processing into the object world:

Implicit simultaneous operation: There are two possibilities when implementing implicit simultaneous operation:

-Passive objects: convert an asynchronous exchange of information into a sequential and synchronous method call or procedure call. Here, parallel processing of objects communicating with each other is very limited.

- Active Objects: Starts a process for each object. This method results in high resource consumption and can therefore only be implemented with a limited number of objects.

· Explicit simultaneous operation: Here, a group of objects (related to objects) or a sequence of processes (related to tasks) is assigned to a process, where the former has been described in A.Coutts, J.M.Edwards paper "Model-driven distributed Systems", IEEE Parallelism, July 1997, pp. 55-63, and the latter in M. Awad, J. Ziegler's paper "Practical Approaches to Designing Parallelism in Object-Oriented Systems", Software Practice and Experience, 1997.9, Vol. 27(9), pp. 1013-1034. By studying the right half of Figure 3 in the above-mentioned Awad/Ziegler paper and Figure 5 in the above-mentioned Coutts/Edwards paper, it can be seen from the interfaces between objects that partially simultaneously represent the interfaces between processes It is pointed out that the communication between objects can be realized not only through synchronous method calling, but also through inter-process communication in the form of asynchronous information forwarding. Specifying this type of communication at the object interface has the disadvantage of greatly increasing the difficulty of reusing and maintaining said objects.

Especially when communicating between different objects - also called instances - in a distributed system, these objects usually have a so-called client/server relationship with each other and are assigned to different processes, so the The solution described above is a very disadvantageous solution in terms of the reusability and maintainability required in this complex system.

It is therefore the task of the present invention to devise a method in a distributed system for transferring information between client instances and server instances assigned to the different processes in such a way that the highest possible reusability can be provided when executing the method performance while simplifying the stated maintainability as much as possible.

This object is solved by the characterizing features given in claim 1 . The subclaims are characterized by further developments of the invention.

According to the invention, this is achieved in that, in a distributed system, in order to transfer information between a client instance assigned to a first process and at least one server instance assigned to at least one other process, additionally A number of partner instances that are set as communication partners are used. After receiving a message pointing to at least one server instance from the client instance, the first instance of the partner instance including the first process selects at least one suitable one from the partner instances, including at least Another instance of another process for message reception and forwarding. Another instance of said at least one other process transmits said information to at least one server instance addressed by it and, if necessary, obtains a message from said at least one server instance in order to pass said information through said at least one server instance A first instance of a process passes the information to said client instance.

In this way, the definition of the communication mode between the client instance and the at least one server instance is transferred to the partner instance which contains a process and is set as the communication partner. Thus, said information is synchronously transferred between a client instance and a first instance comprising said first process, for example by procedure call or method call, and between said at least one server instance and said at least one other between at least one other instance of a process. The transmission of information between the first instance containing the first process and another instance containing at least one other process can then be decoupled synchronously or asynchronously from the communication interfaces of the client instance and the at least one server instance. This advantageously achieves maximum reusability with regard to the implementation of the client instance and the at least one server instance. Also, since the communication interface between said first instance containing the first process and another instance containing at least one other process has to be matched, the communication between said client instance and at least one server instance is not touched interface, so the maintainability of said is also greatly improved.

A further refinement of the invention provides that the first instance containing the first process selects the other instance containing at least one other process by means of an assignment table. The types of information that the client instance can send and the address of the other instance containing at least one other process are entered in the allocation table. The allocation table has the advantage that its content can be changed at any time, and the first instance containing the first process can be quickly selected.

According to an advantageous development of the invention, the selection with the first instance containing the first process can be changed dynamically depending on the system load. In this way, system failures and blockages can be avoided during the allocation of the processes described to the individual processors.

A further refinement of the invention relates to the special case in which the first process coincides with the at least one other process. In this case, the first instance containing said first process and another instance containing said at least one other process are combined into one instance. The invention can thus be applied to this special case without adaptation.

A further advantageous refinement of the invention lies in the manner of implementation. All instances (client instance, server instance, instance containing the first process, partner instance, etc.) can then be implemented in the form of objects whose structure is determined by the object class. In this way, the first instance containing the first process and the further instance containing at least one other process advantageously always have a common object class structure. Using pure object-oriented programming principles in this way, a high degree of modularity, high reusability and maintainability can be achieved.

A further embodiment of the invention consists in the advantageous application of the method according to the invention in a telephone switching system. Thus, all the above-mentioned advantages can also be obtained by contacting the telephone switching system.

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. In the picture:

Fig. 1 shows an example flowchart of the method of the present invention,

Figure 2 shows an embodiment applied in the field of system alarms in telecommunication systems such as telephone switching systems.

See the appendix at the end of this manual for the legends of the drawings.

FIG. 1 shows a flowchart of the information transfer between a client instance assigned to a first process and a server instance assigned to another process. The client instance and the server instance, i.e. the first instance containing the first process and the other instance containing at least one other process, and some actions performed by the server instance are represented by objects with small boxes form shown. The object "client" then corresponds to the client instance, the object "server" corresponds to the server instance, and the object "object handler 1" corresponds to the first Active instance, object "ObjectHandler2" corresponds to another active instance of said partner instance containing said other process, object "Action" corresponds to an action, and object "ConfirmAction" corresponds to an action for all The response to the requested action. Here, the active instance containing the corresponding process is marked by a thick line box. The mode of the action is determined only when the specific object action is called.

When a client requests an action with a response to be performed by the server, for example run the following method:

The client requests an action from the server, upon which a response will be generated. The action is invoked by the client without knowing which process should execute or on which processor platform the action should be executed. The object handler 1 provides the client with the described call procedure "call_action" for this purpose. After invoking the procedure "call_action" - also known as a method in object-oriented programming - a univocal number is assigned to the object handler 1 (obtaining the processing number) and the timer is started (Start timer) which triggers an error handling if no acknowledgment is entered in time. Thereafter, the object processing program 1 searches for a partner instance set as a communication partner, such as the object processing program 2 (finding the target object processing program), wherein the object processing program 2 is determined according to the mode of action Assigned to the action, the object handler 1 then transmits the action request message "action_request" to the object handler 2. The information is received by the object processing program 2, and its communication partner, that is, the address of the object processing program 1 is stored together with the number assigned to the object processing program 1 (storing the communication partner), and then the Describes the process (execution) of an object's action. This object action will then cause said server addressed by the client to perform said action by a procedure call action. After performing the described actions, the server indirectly sends a response back to the client in a similar manner. Subsequent procedure calls, information transfers and actions are then executed by the server in the direction to the client. The procedure call "call_action" clears the address of the communication partner and transmits the action request message "action_request" for reply from object handler 2 to object handler 1, which because of its The assigned number is known by the object handler 2, the assigned number is cleared by the object handler 1 (release processing number) and the timer is stopped (stop timer), in order to transmit the response, the object handler 1 calls The process of "execute object confirmation action" is finally executed by the "object confirmation action" of the client's "confirmation_action" process.

When the client requests the server's action without response, the method of information transmission from the client to the server in operation of the present invention is similar to the above. Cancel the operation steps such as "obtaining the processing number", "starting the timer", "storing the communication partner" and the related steps of answering from the server to the client.

In the case of so-called broadcasting, i.e. one client requests an action from several servers, then there are different possibilities:

- If the server addressed by the client is assigned to a common process, the object handler 1 can transfer the said "action_request" message to the object handler 2, which is responsible for multiple servers Execute this action, or, object handler 1 can also send multiple "action_request" messages to a plurality of object handlers 2 including server processes, and these object handlers 2 prompt the server to execute the described server respectively. described action. It is also possible to combine the two variants described.

- If the server addressed by the client is assigned to a different process, the object handler 1 will send an "action_request" message respectively to the object handler 2 containing said different process, and will always be handled by said object Program 2 prompts said server to perform said action.

A combination of all the above-mentioned possibilities is also conceivable here.

Often multiple actions need to be performed in a distributed system, so obviously each server can also act as a client, and each client can also act as a server, and client functions and server functions can be combined into one object.

The advantageous decoupling of the process interface from the object interface of the client and the server can be seen from the fact that the communication between the client and the server is carried out synchronously by means of procedure calls or method calls, where necessary only object processing The information forwarding between program 1 and object handler 2 is realized asynchronously across process boundaries.

In the special case, for example, that clients and servers located on a common platform may be assigned to the same process, the objects "object handler 1" and "object handler 2" are combined into a single Object. As shown in FIG. 1, the object handler 1 in this case sends said "action_request" message to itself.

Figure 2 shows an embodiment applied in the field of system alarms in telecommunication systems such as telephone switching systems.

In the system alarm, for example, there are the following objects, which act simultaneously as client and server and request different actions from each other. Also, these objects can be on different platforms.

The task of the object "Alarm Balance Monitor" (ABM) is to generate an alarm balance for all alarms of the alarm instance (AMOI) it monitors. To generate this alarm balance, the alarm balance monitor requires at least one so-called SIBS object, which resides on a processor platform and provides the monitor with a collection of information about the monitored alarm instances .

The objects "Invoker", AMOI (Alarm Management Object Instance), SIBS (Site Balance Supply) and ABM (Alarm Balance Monitor) are shown in small boxes. Arrows whose type is not listed in the legends in the appendix are used to indicate the transfer of information between the objects in question, which sometimes goes beyond process limits. Here, the information transmission described corresponds to the information transmission between the client and the server shown in FIG. 1 . The "invoker" object can then, for example, act as a client and the AMOI object as a server. This also applies correspondingly to the remaining objects AMOI and SIBS, as well as SIBS and ABM.

After the system alarm calls "Set_Alarm", for example, the following action process is triggered:

- set_alarm: the monitored alarm instance AMOI gets a new alarm from the caller "caller", checks the parameters that determine the alarm (check_parameters), and then creates a new alarm instance (create_parameters) Included warnings).

- Acknowledgment: Sends an acknowledgment from the AMOI instance to the caller instance after the sysalarm call "set_alarm".

- Balancing SIBS: requesting at least one server object SIBS in order to collect the obtained information required for the described alarm balancing (collecting the alarm status of all relevant AMOIs).

- Balancing ABM: said server object ABM is then requested to collect the information obtained by said at least one SIBS object for alarm balancing (collecting the alarm status of all relevant SIBSs).

Since the action is requested across the process boundary, the information is transmitted from one object to another through the first instance of the initiative and another instance of the initiative, for example through Figure 1 which is not shown in Figure 2 Object handler 1 and object handler 2 in .

The object handler 1 can select the object handler 2 by means of an assignment table. The distribution table is as follows, for example: action object handler answer Setting_Alarm AMOI On the AMOI platform yes balanced SIBS On the SIBS platform no Balance ABM on the main platform no

As soon as a certain action can be executed by different server objects, the allocation of the object handlers 2 can be changed according to the system load. Legends for appendix figures:

Claims (6)

1. In distributed system at the client instance of distributing to first process (client) with distribute to method of transmitting information between at least one server instance (server) of another process at least, wherein, receiving one after described client instance points to the information of at least one server instance, come from described partner instance, to select that at least one is suitable by that first example (object handling program 1) that comprises described first process in the partner instance that is set as the communication partner, comprise another example of another process (object handling program 2) at least, so that carry out message pick-up and forwarding, described another example sends described information at least one server instance by its addressing, and obtain an information from described at least one server instance where necessary, so that send this information to described client instance by described first example that comprises first process.
2. 2. the method for claim 1 is characterized in that:

   Described first example that comprises first process is selected described another example of another process at least that comprises by an allocation table, and described allocation table is meant the distribution between the roughly the same described address that comprises another example of another process at least of the kind of the information that described client instance can send.
3. 3. method as claimed in claim 2 is characterized in that:

   Can dynamically change the selection of being undertaken by first example that comprises first process according to described system loading.
4. 4. the method according to any one of the preceding claims is characterized in that:

   If described first process and described another process at least match, then the synthetic example of first example that comprises described first process and another example set that comprises described another process at least.
5. 5. the method according to any one of the preceding claims is characterized in that:

   All examples realize that with the form of object its structure is determined by object class.
6. 6. the method according to any one of the preceding claims is characterized in that:

   It is applied in the telephone switching system.

Images