RU2702502C1 - Infocommunication system resources control method - Google Patents

Abstract

FIELD: information technology.

SUBSTANCE: invention relates to the field of information technologies and can be used in info-communication systems (ICS). In the info-communication system resources control method, by withdrawing them from a process which has requested system resources from the corresponding service server from the corresponding service server, wherein the exchange command is executed by the service server in several steps: forming channel under control of client of service of transfer and delivery of files of ICS, selection by client of port number on its computer and implementation of opening procedure for its port; sending a port number to a server of a service for transfer and delivery of ICS files, issuing a command to the specified port of the client computer by the server.

EFFECT: high intensity of servicing applications and throughput capacity of services based on the average number of timely operations supported by a server of a typical service of transport TCP connections in case of multiple different resources and incomplete a priori information on resources required for processes.

1 cl, 5 dwg

Description

The invention relates to the field of information technology and can be used in infocommunication systems (IKS).

According to the law “On Information ...” an infocommunication network is a technological system designed to transmit information via communication lines, access to which is carried out using computer technology.

One of the main functions of managing information and communication systems is the function of managing its resources. In the context of resource management, the task of managing the services that provide services to the infocommunication system is one of the most important for ensuring the effective functioning of the system itself. In the real world, the functioning of infocommunication systems is constantly affected by such destabilizing factors as failures, failures, software errors, disturbances in synchronization systems, imperfect architectural and design solutions, conflicts, deadlocks, emergency and fan blackouts, viruses, and “hacker” attacks , Spam, etc.

Thus, a technical problem arises, which consists in the low efficiency of well-known services for managing ICS resources in the presence of many heterogeneous resources that are requested dynamically based on the control of deadlock situations. The way to manage services that provide services to the infocommunication system can be considered as a way to manage ICS resources.

The main services of the services are provided by various basic or special protocols based on the methods of monitoring deadlocks, which are the basis of the so-called prevention of deadlocks while simultaneously executing several software processes. The most common should include methods of interrupting the processes described in Trachtengerz E.A. "Software for parallel processes." M., Science, 1987, 272 pp., Dijkstra E. “Interaction of sequential processes, in the book. Programming Languages ”, M., Mir, 1972, p. 3 ... 86, Haberman A.N. Prevention of System deadlocks. - Communications ACM, 1969, vol. 12, N 7, p. 373 ... 377.

The disadvantage of these methods of interrupting processes is that in the block for generating threshold values, the tolerances for the system parameters are assigned without taking into account its technical condition, loading communication channels and buffer devices of switching nodes.

There are also a number of ways to control deadlocks that occur when several software processes are executed simultaneously. The closest and selected as a prototype of the invention are the methods of interrupting the processes described in patent RU No. 2287220 "System and method for preventing deadlocks using a timer for high-speed downlink packet access", publ. 10/11/2006

A well-known analogue of the invention is “A method for dynamically monitoring deadlocks of an infocommunication system and a device for its implementation” (RF patent RU No. 2502123 dated 12/20/2012, published in Bull. 35). The technical result of this invention is to increase the efficiency of identifying deadlocks, especially in the case of the presence of many different types of resources, with incomplete a priori information about the resources required for the processes, taking into account the attributes of critical resources - reliability indicators of technical, software resources and buffer memory sizes of nodes of the information and communication system. The indicated technical result is achieved by determining the mathematical expectation of the failure rate of the i-th critical technical resource, the size of the buffer memory area of the infocommunication system node, setting the time interval of the planned process execution and calculating the availability coefficient, comparing a certain availability coefficient with a threshold level and making conclusion about the presence of deadlocks in the infocommunication system.

Its undoubted advantage is simplicity. The disadvantage of this method is that it is ineffective if there are many different types of resources that are dynamically requested, it requires frequent removal of processes, which leads to unproductive time costs, as well as low efficiency when architectural deadlocks occur, caused by many different types of critical resources with such attributes, as indicators of the reliability of technical and software resources and the size of the buffer memory of system nodes.

The technical problem solved by the invention is the low efficiency of the known methods for managing ICS resources in the case of the presence of many heterogeneous resources requested dynamically based on the control of deadlock situations.

The technical result of the invention is to expand the capabilities of technical means of managing ICS resources to increase the efficiency of control of technical and software resources in the case of the presence of a multitude of heterogeneous resources and with incomplete a priori information about the resources necessary for the processes, and in general the performance of a computer complex using transmission and delivery services files.

The proposed method is based on such a sequence of interruption of a process that requested some system resources occupied at the time of the request, which interrupts and removes all resources from it, and after releasing the required resources by other processes, the interrupted process is activated again until the next request for occupied resources.

The technical result of the invention is achieved by distributing the various stages of the operation of loading the servers of the ICS services in such a way that the quality function is a linear function of variables when performing the whole complex of unit sub-operations at all stages of operation (9) would go to the maximum.

The proposed method for managing the resources of the infocommunication system for organizing and managing typical ICS services consists in the following sequence of actions:

- determining the location of service nodes on networks and ICS nodes for various types of operations, based on the requirements of sustainable provision of all users with services for the remote receipt, transfer and transfer of files and linking a group of users to the corresponding servers of a typical service;

- the procedure for assigning addresses to users of ICS services;

- calculation of the main server parameters (the number of user and network ports, their throughput, throughput of the binding lines, channels and paths);

- implementation of the management procedure in accordance with the values of the main parameters of the servers.

The main feature of the information or telecommunication services provided by the ICS is that regardless of the type of service, even for those that involve the organization of interaction between two separated users, it is provided with the involvement of the corresponding service server. The nodes of the ICS services are usually located at its border nodes (on the border nodes of the ICS between access networks and the transport network).

The functioning of each type of ICS services is determined by the requirements for the services of this type and the protocols used, which are built according to the standard client-server scheme.

The work of services at the user level involves several stages:

- identification (entering the identifier name and password);

- selection of a catalog;

- determination of the exchange regime;

- execution of exchange commands;

- completion of the procedure.

The service protocol supports two logical connections when organizing an FTP or SMTP session in the ICS between the ICS computers (Fig. 1), one of which is designed for remote access, and the other is for data exchange.

The server performs a port open operation and waits for a connection with the client. The client performs a port activation operation. The channel remains active until the procedure is completed. At the same time, the type of IP service corresponds to a minimum of delay, since this channel is used to manually enter commands. A data channel (TCP or SMTP protocol) is formed each time to send files or letters. The channel opens before the start of the transfer and closes by the code of the end of the file. The type of IP service in this case is oriented towards maximum throughput.

The end user of the ICS interacts with the protocol interpreter, whose tasks include managing the exchange of information between the user and the file system, both local and remote.

First, at the request of the ICS user or application process (service clients), a control channel is formed, which is subsequently used to transmit commands from the client and responses from the server. The information channel is formed by the server at the command of the client, it should not exist continuously throughout the entire FTP or SMTP session and can be formed and liquidated as necessary.

The control channel can be closed only after the completion of information exchange. After the channel is formed, the client can send commands on it. The server perceives them, interprets and transmits responses.

When organizing information exchange in the ICS between two remote FTP or SMTP file exchange servers, when, on the initiative of the ICS user, an exchange is performed between two FTP servers (SMTP) (Fig. 2), any exchange command is performed in several stages:

- channel formation under the control of a client of the ICS file transfer and delivery service;

- the client selects a port number on his computer and performs the “open” procedure for this port;

- sending by the client to the server of the service of transfer and delivery of ICS files a port number using the PORT command;

- server issuing a command to the specified port of the computer client.

The description of the commands of the ICS file transfer and delivery service is given in the standard of the RFC 959 Internet Architecture Council.

Service teams are unified, the user set of commands can vary. The process of functioning of the ICS file transfer and delivery service due to the influence of a number of random factors (the randomness of applications from ICS users or application processes to receive this service, a random number of applications themselves, etc.) is probabilistic.

При получении положительного отклика происходят процессы идентификации, выбора режима и каталога, которые занимают случайное время t_pridn.The user of the ICS, accessing the file transfer and delivery service, expects a response from the nearest FTP or SMTP server of the ICS for a random time

Upon receipt of a positive response, identification, mode and directory selection processes occur that take a random time t _pridn .

The FTP or SMTP client of the service issues the required commands and receives files using the protocol interpreter, which controls the exchange of information between the ICS user and the file system (local and remote) of the file transfer and delivery service or the ICS mailbox system. The time to obtain the required information t_pi also a random variable. It is minimal if files are sent from the local file system, and increases when they are sent over ICS networks. Time t_pi also depends on the size (volume) of the requested file. Thus, the total time of interaction between the user of the ICS and the service is a random variable and is equal to:

.As a rule, users of the ICS client of the file transfer and delivery service are attached to the corresponding service node, which hosts an FTP or SMTP server with finite performance and a limited number of ports - n. They are served by the server, if it has a free resource, or go into standby mode. In accordance with this, the model in the form of a set of queuing systems with expectation, each of which, in accordance with the symbols of Kendall, can be defined as a system, is selected as a mathematical model of a typical service

.

The flow of requirements for services from a group of ICS users is random, ordinary with no aftereffect.

In practice, it is permissible to assume that a real flow has such properties if it is formed by superposition of private flows from independent sources, when their number is more than 8-10. Typically, these conditions are fully satisfied in the ICS.

The main parameter of servicing ICS users is the intensity of service μ, which, to a first approximation, is equal to the inverse of the average value (mathematical expectation) of the total time of interaction of each client with the ICS service, i.e. μ _cc = (t _cc ) ^-1 .

The state of a component of a typical ICS service in time for transferring and delivering files on an access node from the side of a group of ICS users is presented as a graph (Fig. 3).

In FIG. 3 arrows show the intensity of the transition from one state of a component of a typical ICS service to another. Each state of a service component is characterized by the number of occupied ports of the service server at the application level. Then, from the side of the ICS user group, each component of the ICS service is dynamically described by a system of stochastic differential equations that specify the probabilities of the corresponding states:

In stable situations, the process of functioning of the component of the ICS service will go into some steady state, characterized by the final state probabilities. Moreover, the probabilities that all ports of the server of the ICS service component are free, exactly k ports are occupied, all ports of the server are busy providing the services of the service and s of the ICS users are waiting for service, are respectively

Where:

Usually, in any typical ICS service, a time limit for waiting for service to start service is introduced (for example, the waiting time should not exceed t _max ), then, obviously, this can only be guaranteed with some probability equal to

Where

Under these conditions, during the operation of the ICS, it is important to ensure a given probability that the allowable waiting time for the service will not exceed the value calculated on the basis of relation (6). At the same time, for the given probability values (6) and the maximum allowable value of the service waiting time, control parameters are determined that specify, for the necessary exchange traffic, the quantitative values of the service parameters of the paired components of the ICS service: the number of allocated ports for each server of the typical service assigned to a certain group ICS users (KSA application processes) and the acceptable bandwidth value of supported secure TCP transport connections for a typical service. At the same time, the number and throughput of supported TCP transport connections are not a limitation for the ICS service.

For certain groups of ICS users, a more important indicator of the provided services of the ICS service is the average waiting time for the start of servicing each requirement. In this case, control is carried out according to the criterion of providing a given value of the average waiting time, not exceeding a certain value:

In accordance with a certain value of time T _gm , beyond which it will be considered that the ICS service is not efficient enough and efficient, the number of allocated ports for each server of the typical service assigned to a specific group of ICS users and the allowable value of the throughput of supported secure transport TCP connections are calculated .

When implementing the proposed method, the number and throughput supported by the server of a typical service of transport TCP connections are not a limitation for the ICS service.

. При этом время t_cc определяется в основном временем t_pi, а интенсивность обслуживания определяется выражением

.In most practical cases of the operation of the ICS, the volume of the transferred files or letters is such that the time

. Moreover, the time t _cc is determined mainly by the time t _pi , and the service intensity is determined by the expression

.

The mathematical expectation of the random time t _pi is estimated as follows. In a first approximation, the time for sending a file or letter in the ICS service is determined by the time it takes to search for the required document in the database or the t _poisk file system, its volume and effective speed of file transfer over the established TCP transport connection. The average time t _{pi is} determined from the expression:

In expression (8), the quantities and probabilities with the index y mean the values of the corresponding quantity of the jth type, Q _j is the average file size (letters, service messages) of the jth type, V _eff (j) is the effective transmission rate according to the established j- MCP connection.

Thus, in the claimed method, when organizing and managing typical ICS services, the following actions are performed (Fig. 4):

- the location of service nodes on the ICS networks and nodes is determined for various types of operations, based on the requirements for the sustainable provision of all ICS users with the services of remotely receiving, transmitting and forwarding files or letters, and linking user groups to the corresponding servers of a typical service;

- The procedures for assigning addresses to users of the ICS service are being carried out;

- the main parameters of the hosted servers are calculated (the number of user and network ports, their throughput, throughput of the binding lines, channels and paths);

- management procedures are carried out in accordance with the values obtained.

The solution of the ICS resource management problem within the framework of the ICS service management tasks for two main services: the ES service (e-mail) and the file exchange service (FS) is carried out as part of the linear programming model as follows.

Suppose that in the IKS there are deployed an ES service that has N _ES servers, and a FD service that has N _FD servers.

Services participate in the operation, consisting of four stages O _VKS1 , O _VKS2 , O _VKS3 , O _VKS4 , and each server can provide any of the four stages of the operation, but, depending on its location on certain ICS nodes, to a different extent .

One ES server can provide a certain share of support for each stage of the operation, expressed in single sub-operations: for the first stage of operation O _VKS1 -Serv _ES (O _VKS1 ), for the second stage of operation O _VKS2 -Serv _ES (O _VKS2 ), for the third stage of operation O _VKS3 -Serv _VC (O _VKS3) and for the fourth stage of operation O _VKS4 -Serv _VC (O _VKS4).

Similarly, one FD server can provide a certain amount of support for each step of the operation: for operation O _VKS1 -Serv _FD (O _VKS1 ), for operation O _VKS2 -Serv _FD (O _VKS2 ), for operation O _VKS3 -Serv _FD (O _VKS3 ) and operation for O _VKS4 -Serv _FD (O _VKS4).

Each single suboperation of each stage provides a certain contribution to a certain indicator of the quality of the operation, moreover, a single suboperation of the first stage of the operation provides a unit quality of ₁ , a single suboperation of the second stage of the operation provides a unit quality of ₂ , a single suboperation of the third stage of the operation provides a unit quality of ₃ and unit the suboperation of the fourth stage of the operation provides a single quality with ₄ .

For the operation to be successful, it is necessary that at each stage it is completed: at the first stage, at least n ₁ unit sub-operations of the first stage must be carried out, at the second stage at least n ₂ unit sub-operations of the second stage must be performed, at the third stage it must be carried out at least n ₃ unit suboperations of the third stage and at the fourth stage at least n ₄ unit suboperations of the fourth stage should be carried out.

When managing services, it is necessary to distribute the load of ICS service servers in sub-operations of different stages of the operation so that the quality of its execution is maximum.

- число серверов ЭП, занятых осуществлением поддержки выполнения единичных подопераций первого этапа операции,

- число серверов ЭП, занятых осуществлением поддержки выполнения единичных подопераций второго этапа операции,

- число серверов ЭП, занятых осуществлением поддержки выполнения единичных подопераций третьего этапа операции,

- число серверов ЭП, занятых осуществлением поддержки выполнения единичных подопераций четвертого этапа операции,

- число серверов ФО, занятых осуществлением поддержки выполнения единичных подопераций первого этапа операции,

- число серверов ФО, занятых осуществлением поддержки выполнения единичных подопераций второго этапа операции,

- число серверов ФО, занятых осуществлением поддержки выполнения единичных подопераций третьего этапа операции,

- число серверов ФО, занятых осуществлением поддержки выполнения единичных подопераций четвертого этапа операции.To solve this problem of managing ICS services, we formalize this task. To do this, we introduce the following notation:

- the number of ES servers engaged in supporting the implementation of individual sub-operations of the first stage of the operation,

- the number of ES servers engaged in supporting the implementation of single sub-operations of the second stage of the operation,

- the number of ES servers engaged in supporting the implementation of single sub-operations of the third stage of the operation,

- the number of ES servers engaged in supporting the implementation of single sub-operations of the fourth stage of the operation,

- the number of FD servers engaged in supporting the implementation of individual sub-operations of the first stage of the operation,

- the number of FD servers engaged in supporting the implementation of single sub-operations of the second stage of the operation,

- the number of FD servers engaged in supporting the implementation of individual sub-operations of the third stage of the operation,

- the number of FD servers engaged in supporting the implementation of single sub-operations of the fourth stage of the operation.

,

,

,

,

,

,

,

, которые задаются так, чтобы качество выполнения операции было максимально.Thus, eight variables (solution elements) appear in the control problem:

,

,

,

,

,

,

,

which are set so that the quality of the operation is as high as possible.

When performing the whole complex of individual suboperations at all stages of the operation, the quality indicator has the form:

,

,

,

,

,

,

,

,

(х - число серверов, занятых осуществлением поддержки выполнения единичных подопераций первого этапа операции), чтобы функция (9) от них обращалась в максимум.When organizing the management of ICS services, such non-negative variable values are selected

,

,

,

,

,

,

,

,

(x is the number of servers engaged in supporting the implementation of single sub-operations of the first stage of the operation) so that function (9) addresses them to the maximum.

In this case, the following restrictive conditions must be fulfilled related to the finite number of servers and the required number of unitary sub-operations performed, i.e.:

where: Serv _EP (About _BKS1 ) - the share of support for each stage of the operation, expressed in single suboperations;

N _EP - the number of deployed email service servers;

N _fo is the number of file sharing service servers deployed.

,

,

,

,

,

,

,

, удовлетворяющие линейным неравенствам (10) и (11), при которых функция качества - линейная функция этих переменных (9) обращается в максимум.Thus, the final task of managing ICS services in a developed way is formulated as follows: in the process of ICS functioning, when performing a certain operation, the service resources are distributed, the corresponding service servers are allocated for performing single suboperations, i.e. variable values are set

,

,

,

,

,

,

,

satisfying the linear inequalities (10) and (11), for which the quality function - the linear function of these variables (9) turns to the maximum.

A schematically developed method for managing the resources of ICS services during an operation is shown in FIG. four.

Comparison of the developed method of resource management services by the average number of timely operations with the existing method was carried out according to the method described in the journal “T-Comm. Telecommunications and Transport ”, Volume 11, No. 2 of 2017, p. 48 ... 50. The comparison results are shown in FIG. 5.

Comparison results of the developed method for managing the resources of ICS services (Fig. 5) show a significant increase in qualitative indicators (2.1-3.2 times on average, depending on the number of servers in the services) compared to existing methods for the non-operational allocation of resources.

Claims (6)

A method for managing the resources of an infocommunication system by removing them from a process that requested service resources occupied at the time of the request from the corresponding server for the system for waiting time and transferring them to the current process that needs resources, which differs from previously known resource management functions when organizing a file transfer and delivery service the fact that the proposed method is based on such a sequence of interruption of a process that requested some system resources occupied at the time of the request from the service server: the client performs the port activation operation, the channel remains active until the procedure is completed, while the type of IP service corresponds to a minimum of delay, since this channel is used to manually enter commands, the channel opens before the transfer starts and closes by the end-of-file code, when This exchange command is executed by the service server in several stages: - channel formation under the control of a client of the ICS file transfer and delivery service; - the client selects a port number on his computer and performs the “open” procedure for this port; - sending by the client to the server of the service of transfer and delivery of ICS files a port number using the PORT command; - server issuing a command to the specified port of the computer client, at the same time, the ICS service nodes are located at its border nodes (on the ICS border nodes between the access networks and the transport network), which interrupts and removes all resources from it, and after releasing the required resources by other processes, the interrupted process is activated again until the next request for occupied resources, which increases the request servicing intensity and the throughput of services by the average number of timely operations supported by the server of a typical service of TCP transport connections in the case of multiple heterogeneous resources and with incomplete a priori information about the resources necessary for the processes.

Images