Collaborative optimisation method, device and system. RU patent 2520354.

FIELD: radio engineering, communication.

SUBSTANCE: invention relates to communication engineering. The method involves: obtaining link information, server and bandwidth information, information on user requirements in a subnetwork, where the server and bandwidth information includes virtual server bandwidth of each external port of the subnetwork, and the virtual server bandwidth is server bandwidth outside the subnetwork, where the server bandwidth outside the subnetwork is required for the subnetwork through an external port; obtaining an optimum routing parameter and a subnetwork server selection parameter in accordance with link information, server and bandwidth information and information on user requirements; obtaining optimised input bandwidth of each subnetwork external port in accordance with the optimum routing parameter and server selection parameter; comparing the optimised input bandwidth and the virtual server bandwidth of each external port, and if the results of comparing optimised input bandwidth and virtual server bandwidth of all external port are less than a set error value, the optimum routing parameter and the server selection parameter are used in the subnetwork.

EFFECT: improved performance of the entire network.

14 cl, 6 dwg

The technical field

The present invention relates to the field of communication technologies, and more specifically to the way, device and system for joint optimization.

The level of technology

Task ISP (Internet Service Provider, the provider (ISP) services the Internet, in General, is to ensure access to the Internet and needs to deal with the regulation of traffic (I.E. Traffic Engineering), that is to determine the optimal route for the data stream to minimize network congestion. Task CF (Content Provider, the provider of online information) is to supply the required information to the user, and require the solving of the tasks of the server selection (SS, Server Selection), that is an indication of the optimal server for different users to minimize end-to-end delay. When the ISP and CF perform, in an appropriate manner, and THOSE SS, this is not an optimal state from the point of view of the entire network.

For optimum performance throughout the network, at the moment, begin to attempt to perform joint optimization (Joint Optimization) using solutions Nash (NBS, Nash Bargain Solution) and convex optimization of game theory to implement cooperative games for THOSE specified and SS, who are the two optimization problems, in order to achieve overall optimum balance.

Mathematical model of joint optimization and THOSE SS performed using the NBS, is the task of the limited convex optimization. On prior art exists optimization method based on centralization. In this scheme suggest that a computing device collects information about all the network required for solving optimization problems, applies the method of centralization for optimal solutions for system and applies this best solution to the network to complete change policy routing and selection policies of the new server. This pattern, however, is difficult extensibility. When you zoom network time spent on gathering information and calculations, is growing rapidly. If the joint optimization is applied to the network of city-wide or even the whole country, the use in real time of an existing way of joint optimization of the network it is difficult to guarantee.

Summary of the invention

In variants of the implementation of the present invention, a method, device, and system that effectively reduce the amount of computation and complexity of the joint optimize and improve the performance and efficiency of the entire network.

The way of joint optimization includes:

obtaining information about the line of communications, information about the server and bandwidth requirements information user subnet, where information about the server and bandwidth includes throughput virtual server for each external port subnet, and bandwidth virtual server is the bandwidth of the server on the subnet, and the throughput of the server on the subnet is required for the subnet via external port;

getting parameter optimal routing and select parameter server subnet in accordance with the information about the line of communications, information about the server and bandwidth, as well as information about the requirements of the user;

receive optimized input bandwidth for each external port subnet in accordance with the option of optimal routing and selection of the server; and

comparing optimized input bandwidth and throughput virtual server for each external port, and if the comparison is optimized input bandwidth and throughput virtual server for all ports less than the threshold error, the application of parameter optimal routing and select parameter server on the subnet.

In one embodiment, the present invention proposed computing device for joint optimization, including:

collection module settings made with the possibility of receiving information about the line of communications, information about the server and bandwidth, and information about the requirements of the user in the subnet where information about the server and bandwidth includes information about the capacity of a virtual server for each external port subnet and bandwidth virtual server is the bandwidth of the virtual server on the subnet, and where server bandwidth the subnet is one that requires a subnet via external port;

module of calculations performed with the option of optimal routing and select parameter server subnet in accordance with the information about the line of communications, information about the server and bandwidth, and information about the requirements of the user, and an optimized input bandwidth each external port subnet in accordance with the option of optimal routing and selection of the server; and

output module, made with the possibility of comparison optimized input bandwidth and throughput virtual server for each external port, and if the comparison is optimized input bandwidth and server capacity for all external ports less than the threshold error, the application of parameter optimal routing and select parameter server on the subnet, and the output parameter optimal routing and select parameter server.

In one of the embodiments of the present invention proposed network communication, including:

system to obtain joint optimization of information on-line, information about the server and bandwidth, and information about the requirements of the user in the subnet where information about the server and bandwidth includes throughput virtual server for each external port subnet.

calculation of the parameter optimal routing and select parameter server subnet in accordance with the information about the line of communications, information about the server and bandwidth, and information about the requirements of the user, and the optimised input bandwidth the ability of each external port subnet; and

comparing optimized input bandwidth of each external ports and bandwidth of the external port of the virtual server, and the result of the comparison is less than the adjusted value of the error, use the optimal routing and select parameter server for the network.

In one of the embodiments of the present invention of the proposed system for joint optimization, including:

a computer device joint optimization made with the possibility of collecting information on-line, information about the server and bandwidth and information about the requirements of the user receiving the routing option and select parameter server and an optimized input bandwidth and optimized output bandwidth each port; and

the routing engine content, made with the possibility of conversion of routing option and select parameter server received computing device for the joint optimization of the routing option and the option to select a server, primenyaemye in the local network.

It is obvious, that the technical solutions proposed in the variants of the implementation of the present invention, the above options to the implementation of the present invention calculations for joint optimization execute in parallel for each subnet. The server on the subnet, refers to the virtual server on the device at the entrance subnet bandwidth virtual server is the lower of the values obtained by subtracting the input background traffic from optimized output bandwidth, calculated with the joint optimization for that port peer-to-peer subnet and the values, obtained by subtraction of the input background traffic from the input capacity of the port subnet minus input background traffic, and through interaction between bandwidth virtual server, i.e. bandwidth subnet, and the calculation of joint optimization performed repeatedly on each subnet in order to get a policy for the selection of route and selection policy server on the network, and apply this policy routing and selection policy server on the network to optimize performance and utilization.

Brief description of drawings

For clearer illustration of technical solutions according to the options of the implementation of the present invention below is a brief description of the accompanying drawings. Obviously, accompanying the drawings in the following description applies only to some variants of the implementation of the present invention and specialists in the art can easily get other drawings based on the accompanying drawings.

Figure 1 shows an implementation option, in which the network to perform joint optimization is divided into two levels;

figure 2 shows an implementation option, in which the network to perform joint optimization is divided into three levels;

figure 3 shows an implementation option computing devices together optimization;

figure 4 shows a two-tier partitioning mathematical model of joint optimization;

figure 5 illustrates the case of the implementation process of the implementation of the joint optimization for the whole network; and

figure 6 illustrates the case of the implementation of the management system for joint optimization.

The detailed description of options for the implementation of

Figure 1 shows a network for execution of joint optimization, broken on two levels, where the first level is the level of access. BAS (Broadband Access Server, Broadcast access server) taken as a structural element for the separation of the network, that is, BAS, switch and DSLAM (Digital Subscriber Line Access Multiplexer, the Hub of the digital subscriber line access), United with BAS, make subnet, and BAS border router (BR, Border Router) and the main router (CR, Core Router), United described above, constitute subnet top level. On the access level, the use of BAS as a structural element of separation subnet is a way to separate subnet, and the border router or other device can be used as a structural element to separate subnet.

Figure 2 shows one of the ways in which the network to perform joint optimization is divided into three levels. The first level is the level of access that is divided into multiple subnets, in which BAS and border router are the structural elements, the average level is divided into different subnet in the regions as structural elements, and the upper level is a subnet, which includes the networks node, for example, the output router.

It should be clear that the network is not limited to split into two or three levels and can be divided into multiple levels. When splitting on many levels, from the bottom up can be consistently named as the subnet of the first level, subnet second level, subnet third level... and subnet level Meters, and the network of higher level can also be called a subnet top level, and the subnet of the first level may also be called a subnet lower level.

Figure 3 shows an implementation option computing devices together optimization. This computing device for joint optimization involves the collection module parameters, the calculation engine and an output module.

Collection module settings is responsible for gathering information about the network status information about the server, information about the user requirements and optimized output bandwidth received from neighbouring subnet, and transmits the input information for the calculation engine. Information about the state of the network includes the bandwidth of the connection line, quality of service (QoS) for communication lines and the cost of transfer on the communication line. Information about the requirements of the user includes a set of users and user requirements to the content. Server information includes a set of servers and server throughput. The set of servers that includes a server on the subnet and virtual server at the entrance of the neighboring network. Optimized output bandwidth output neighbouring subnet is output bandwidth, which can be obtained after the calculation is complete joint optimization in neighboring subnet, and this bandwidth is used as the capacity of a virtual server on the entrance neighboring subnet.

The calculation engine receives information about the status of the network, server information, user information and optimized output bandwidth output neighbouring subnet, which gathers a collection module parameters and calculation engine performs calculations in accordance with the algorithm of joint optimization to calculate the output of bandwidth that can provide subnet required input bandwidth, and the appropriate policy routing and selection policy server, and passes the results to the output module.

Output module accepts optimal results, calculated by the calculation compares the input bandwidth required for this subnet, and output network bandwidth that can provide neighboring subnet, and if the difference between two of them is within the error range, the output module sends routing policy and the policy server selection mechanism routing content. At the same time, the output module sends output network bandwidth that can provide the subnet on the routing of content.

For additional understanding of the principle of realization of the above devices together optimization, the principle of the virtual server on the entrance neighboring subnet and the principle of creating mathematical models of joint optimization will be further illustrated below.

The user subnet can download content from a server on the same subnet, and can download content from a server on another subnet or server subnet top level. When the user subnet queries the server on the subnet, the request must go through the exit, which joins this subnet is another subnet. Therefore, taking subnet as a structural element, the server on the subnet can be considered as virtual server located on the device at the entrance of a given subnet, and the least amount of bandwidth that can provide all servers neighboring subnet maximum output bandwidth neighboring subnet and maximum input bandwidth of a given subnet take as bandwidth that can provide a virtual server on the input given subnet. The smallest amount of bandwidth that can provide all servers neighboring subnet and maximum output bandwidth neighboring subnet is also called optimized output bandwidth neighboring subnet. If not specified, the maximum output bandwidth subnet is the value obtained by subtracting the output of background traffic from the output bandwidth a given subnet, and the maximum input bandwidth subnet is the value obtained by subtracting the output of background traffic from the input bandwidth of a given subnet.

For this subnet assume that the set T users is the set of all users requesting the content of the SR in this subnet, the set's servers is the set of all servers on the same subnet and virtual server on the device at the entrance. Throughput of the server in the subnet is bandwidth, which can provide the server and bandwidth virtual server on the device at the input is some value that is not greater than the value obtained by subtracting the input background traffic from the input bandwidth subnet. When the initial calculation, the value can be predicted on the basis of statistics in accordance with the information about the network. After the withdrawal of the neighboring network optimized output bandwidth, throughput virtual server on the device at the entrance is the lower of the values obtained by subtracting the input background traffic from optimized output bandwidth, derived neighbouring subnet, and the values obtained by subtracting the input background traffic from the input bandwidth of a given subnet. A set I is the initial sites is a set of nodes requesting background traffic on the subnet, and the set J end nodes is set output nodes of the regional network, and, from the point of view subnet suggest that there is a virtual end node background traffic on the device at the entrance.

In this subnet, the model algorithm of joint optimization based on the use of technology convex optimization and NBS (Nash Bargaining Solution, the solution NesA) from game theory, that is the task of the limited joint optimization set when the account restrictive conditions, such as information on-line, information about the server and bandwidth, and information about the requirements of the user. The problem of optimization decide to obtain optimal set of values and a set of optimal values reflect policy changes optimal route THOSE policies and select the optimal server SS. In accordance with this set of optimal values can be calculated user requirement for server bandwidth, and the requirement also includes the requirement for bandwidth on the server on the subnet, i.e. the demand for bandwidth virtual server on the device at the entrance. The lower of the values obtained by subtracting the background traffic from the required bandwidth, and values, obtained by subtraction of the background traffic from maximum input throughput of this port subnet take as optimized input throughput of this port subnet. In accordance with this set of optimal values can be additionally calculated remaining bandwidth that a server can provide to the outside world through the port subnet. The smallest value from the value calculated by subtracting the background traffic from bandwidth and values, obtained by subtraction of the background traffic from maximum output throughput of this port subnet take as optimized output throughput of this port subnet.

Figure 4 shows a variant of realization of the decision of the joint optimization. After the first break level of the initial tasks of joint optimization, the two subtasks, SS-NBS and TE-NBS, optionally split at the second level from the point of view of communication lines, servers and endpoints background traffic to split the task into smaller solved subproblems that, therefore, reduces the execution time of the algorithm.

Splitting principle is the following.

First, on the first level, a mathematical model of the joint network optimization break in accordance with the method of pair split to split into two subtasks SS-NBS and TE-NBS, as well as basic task Dual TE-SS , and the main task Dual TE-SS manages two subtasks SS-NBS and TE-NBS through variables λ l , a u l , and l V rates that are introduced by way of a split. Because the two subtasks SS-NBS and TE-NBS still are complex optimization problems, you must perform additional break. In this case, choose line 1 connection and the server's as structural elements, on the second level sub-task SS-NBL divided into two subtasks SS-NBL-L and SS-NBL-S, as well as basic task DualSS-NBS, and the main task DualSS-NBS manages subtasks SS-NBL-L and SS-NBL-S through the variables p l γ s , and n t rates enacted at the second level of disaggregation. Because the two subtasks SS-NBL-L and SS-NBL-S are the least solved subproblems, additional splitting two subtasks is not necessary. Otherwise, the same way of splitting still need to be adapted for additional splits to get the lowest solved subproblems. Similarly, taking the communication line 1 and end point j background traffic as structural elements, on the second level sub-task TE-NBS divided into two subtasks TE-NBS-L and TE-NBS-J, and the main task DualTE-NBS, and the main task DualTE-NBS manages subtasks TE-NBS-L and TE-NBS-J variables from EQ l and t ij rates enacted at the second level of disaggregation. Because the two subtasks TE-NBL-L and TE-NBL-J are the least solved subproblems, additional splitting two subtasks is not necessary. Otherwise, the same way of splitting still need to be adapted for additional splits to get the lowest solved subproblems.

The above process-level partitioning will be further detailed below.

1: the Initial task of the joint optimization

Using the convex optimization and NBS in game theory, mathematical model of the problem of joint optimization in the subnet set following:

maximize

provided that

integral

integral

integral

integral

integral

integral

integral

integral

variables

integral

In this formula (THOSE 0 , SS 0 ) represent the current state, there is an optimal values obtained when THOSE and SS optimized independently; g l (·) and h l (·) are functions of the joint optimization received THOSE and SS respectively, and both are functions of line 1 of communication;

are, respectively, background traffic lines 1 connection and traffic of the first selected WED line 1 communication from the point of view of the ISP;

represent, respectively, the traffic CF communication lines 1 and the first selected background traffic lines 1 communication from the point CF;

is the traffic generated by a pair (s, t) line 1 communication; x ij and

are, respectively, background traffic pair (i, j) and the proportion of background traffic generated by a pair (i, j) line 1 connection; l , M, t , and B s are, respectively, the capacity of communication lines 1, the user's requirements t and server throughput s; I V=j , is an indicator function that means that the function takes the value 1 when the node v is a node j, and the function takes a value of 0 when the node v is a node other than the node i and node j; meaning I V=t I V=s is the same as for I, V=j ; l is given In(V) and l is given Out(V) represent, respectively, the line of communication through which traffic enters the node v, and the line of communication through which traffic goes from the node v. Variables

are optimal variables that need to be solved in the joint task of optimization.

Perform conversion Lagrangian for the part above formula to obtain:

When using the method of pair split, the above formula can be divided into two subtasks SS-NBS and TE-NBS, and two sub-tasks are managed by the main task Dual TE-SS (? l , a u l , V (l ).

(1.1): subtask S-NBS

maximize

provided that

integral

integral

integral

integral

integral

variables

integral

(1.2): subtask TE-NBS

maximize

provided that

integral

integral

integral

variables

integral

(1.3): Basic subtask Dual TE-SS (? l , a u l , V, l )

Create H (? l , a u l , V (l ) in the maximum value of the objective function subtasks SS-NBS, and G (? l , a u l , V (l ) as the maximum value of the objective function subtasks TE-NBS, where

The main task can be presented as:

to minimize

variables

In the analysis of subtasks SS-NBS and subtasks TE-NBS was found to be two problems can be further divided from the standpoint of communication lines, servers and endpoints background traffic, so to make them broken into smaller tasks that can be solved in parallel.

2: Additional split subtasks SS-NBS

Subtask SS-NBS

in (1.1) can be expressed as:

maximize

provided that

integral

integral

integral

integral

variables

integral

where T(s) is a set of users served by the server s, and S(t) is a set of servers for some user t. Last restrictive condition means that for subassembly v, which is neither a server nor the user that the traffic coming to the site v, equivalent to traffic coming from the node v. That is a smart host v satisfies the law of traffic exchange. This formula is clearly true, and, therefore, does not matter when it was generated Lagrange expression.

Perform conversion Lagrangian for the above formula to obtain:

In this case l belongs to L(s, t) is the line of communication from the server's up to the user t.

Subtask SS-NBS broken, in line with 1 connection and the server's, two types of subtasks SS-NBS-L and SS-NBS-S. the two subtasks are managed by the primary task

Dual SS-NBS (p l γ l , n (l ).

(2.1): subtask SS-NBS-L:

Line 1 connection

maximize

variables

(2.2): subtask SS-NBS-S:

The server's

maximize

variables

(2.3): the Main objective SS-NBS: Dual SS-NBS (p l γ s , n t )

Create L SS-NBS-L (p, l ) the maximum the value of the objective function subtasks SS-NBS-L, and S SS-NBS-S (p, l , γ s , n t ) as the maximum value of the objective function subtasks SS-NBS-S, where

The main task can be represented as:

to minimize

variables

3: Additional split subtasks TE-NBS

Subtask TE-NBS

in (1.2) can be expressed as:

maximize

provided that

integral

integral

integral

integral

variables

integral

Last restrictive condition means that for subassembly v, which is neither a node i, nor node j, background traffic coming to the site v, equivalent to traffic coming from the node v, that is a smart host v satisfies the law of traffic exchange. This formula is clearly true, and, therefore, does not matter when it was generated Lagrange expression.

Perform conversion Lagrangian for the above formula to obtain:

In this case, l belongs to L(i,j) is the line of communication from I to j.

Subtask TE-NBS broken, in accordance with the communication line 1 and end point j background traffic on two types of subtasks TE-NBS-L and TE-NBS-J. the two subtasks are controlled by the main task Dual TE-NBS (z l , t ij ).

(3.1): subtask TE-NBS-L:

Line 1 connection

maximize

variables

(3.2): subtask TE-NBS-J:

Endpoint j background traffic

maximize

variables

(3.3): the Main objective TE-NBS: Dual TE-NBS (z l , t ij )

Create a L TE-NBS-L (z l ) as the maximum value of the objective function subtasks TE-NBS-L, and L TE-NBS-J (z l , t ij ) as the maximum value of the objective function subtasks TE-NBS-J, where

The main task can be presented as:

to minimize

variables

The calculation engine breaks down, in accordance with the above process of breaking, initial problem into two subproblems with less computing, calculates optimal result and sends it to the output module. Output module outputs the result of the calculation on the routing of content, and the use of the network can change the routing policy and the policy server selection. Change policy route optimally selects traffic for each route, deciding thus, the task of designing traffic. Selection policy server operates on the server (including virtual server subnet) and specifies the bandwidth, which can be a dedicated server, and a new set of customers served by the server.

Figure 5 shows the process of implementation of the joint optimization for the entire network. The network, which should be subjected joint optimization, divided into multiple subnets, and the task of the joint optimization decide for each subnet. If each subnet receives the values optimized output bandwidth from neighboring network in the first decision, the capacity of the virtual server on the device at the input set as the initial value, and the initial value is a value that is not greater than the value obtained by subtracting the input background traffic from the input bandwidth subnet. The value can be predicted on the basis of statistics in accordance with the information about the statistics of the network, and can also be a value randomly chosen from the range, for example, half of the values obtained by subtracting the input background traffic from the input bandwidth subnet. If optimized output bandwidth was received from neighboring network, a smaller value from the value calculated by subtracting the input background traffic from optimized output bandwidth neighboring network, and the values obtained by subtracting the input background traffic from the input bandwidth of the neighboring network, choose as the capacity of a virtual server on the device at the entrance.

The above joint optimization of re-perform on each subnet, and final objective of optimization of all subnets and subnet top level is reduced to the optimal value to achieve a Pareto optimization. Policy routing changes and selection policy server after convergence system send assosiated network elements in the network.

The whole process is implemented as follows.

Step 1: Expected primary input bandwidth is set for each subnet in the network of lower level collecting module parameters collects information about the state of the network (for example, information about communication lines), information about the server and bandwidth, and information about the requirements of the user in the subnet, and the calculation engine calculates the optimal parameter routing and server option subnet in accordance with the partition method, shown in figure 4. Get optimized output network bandwidth that can provide the server where the optimized output bandwidth refers to the smallest of the maximum output capacity of the port and the maximum bandwidth that can provide the server in the subnet. Get optimized input bandwidth subnet where the input bandwidth is the bandwidth that must be provided by the server on the subnet.

Step 2: If there is a subnet of the intermediate level, which is not the highest level, the collection engine system parameters for joint optimization in the subnet intermediate level collects optimized output bandwidth calculated neighbouring subnet lower level, and state information network (for example, information about communication lines), information about the server and bandwidth, and information about the requirements of the user in the subnet of the intermediate level, and sets the estimated input bandwidth of the interface between the subnet intermediate level and subnet higher level. The calculation engine calculates the optimal parameter routing and server option subnet intermediate level in accordance with the partition method, shown in figure 4. Get optimized output bandwidth provided by the subnet to subnet in the network of high level and subnets in the network of the lower level, where optimized output bandwidth refers to the smallest of the maximum output capacity of the port and the maximum bandwidth that can provide the server in the subnet. Get optimized input bandwidth subnet where the input bandwidth is the bandwidth that must be provided by the server on the subnet.

Step 3: For the subnet of the highest level, the collection engine settings a framework for joint optimization in the subnet collects optimized output, capacity, calculated by the neighboring subnet lower level, and state information network (for example, information about communication lines), information about the server and bandwidth, and information about the capacity subnet. The calculation engine calculates the optimal parameter routing and server option subnet in accordance with the partition method, shown in figure 4. Get optimized output bandwidth provided by the subnet for low-level network, where optimized output bandwidth refers to the smallest of the maximum output capacity of the port and the maximum bandwidth that can provide the server in the subnet. Get optimized input bandwidth subnet where the input bandwidth is the bandwidth that must be provided by the server on the subnet.

Step 4: the Estimated initial input bandwidth, established at stage 2, replace optimized output bandwidth of the interface to the subnet of the lower level, calculated in step 3, do step 2, the estimated initial input bandwidth installed in step 1, replace optimized output bandwidth of the interface to the subnet of the lower level, calculated in step 2 : perform step 1, and if the calculated expected input bandwidth and optimized input bandwidth are in the range of error, the parameter optimal routing and server parameter output for use in the appropriate subnet. Otherwise, stage 1, stage 2, stage 3 and stage 4 execute repeatedly.

Proxy module (Proxy)Is a proxy for the user and establishes a bridge between the user terminal and the system for joint optimization, and can also be used to collect query from the user. Proxy module can be placed on the edge of the network, for example posted on the DSLAM.

The routing engine content (Content Routing Engine): Specifies how to get the desired content. When the routing engine content finds that the requested content is not in the local device buffer, it uses the name resolution mechanism (Name Resolution Engine) for a list of sites that contain content, and takes a list of nodes in the quality of information about the server and outputs it to the collection module parameters. The routing engine content gets from the output module computing devices together optimization, policy routing changes and selection policy server, and converts them to change policy route and selection policy server, which can be realized and put into effect, where a policy route changes introduced in the transmission mechanism/redirect to perform routing policy on the lower level, and the server selection policy introduced in module support topology to perform server selection at the level of contents.

The name resolution mechanism (Name Resolution Engine): basically returns a list of matching nodes that contain content that is required by the user.

The adapter GET/PUT content: Retrieves the contents required by routing of content from a device to a local buffer and returns the contents of the routing mechanism of contents so the routing engine content sent the content of the target user.

Unit statistics traffic Statistics Collector): Collects information on the background traffic, information about communication lines etc, and gives you detailed information collection engine parameters.

Unit statistics of Request Statistics Collector): Collects requirements of the user and sends the collected user requirements, the collection module parameter.

System for joint optimization peer subnet: Gets optimized output bandwidth calculated computing device for the joint optimization of a given subnet, peer-to-peer receives, through optimized output bandwidth, the estimated input bandwidth required to carry out joint optimization, and this capacity take as bandwidth virtual server on the device at the entrance.

Collection module parameters: basically collects user requests, information about the servers list, the dynamic characteristics of traffic and optimized output bandwidth, peer-to-peer subnet.

The calculation engine: Uses theory split optimization for calculating the optimal pair (user, server) and route point-to-point generated by each pair (user, server, and speed of distribution of traffic for each route.

Output module: the Module output is responsible for reception of optimum result of a calculation from the calculation engine and enter the result in the routing of content.

The mechanism of transfer/call forwarding: Takes instructions from the routing engine content IP layer performs traffic and introduces the results in the routing of content.

Support module topology: Takes instructions from the routing mechanism of contents at the application level, performs the select server and implements the routing on the level of content.

Interactive system for joint optimization is the following:

1: proxy Module completes the collection of requests content from the user.

2: the name resolution Mechanism resolves content in accordance with the name of the requested content for information about the requirements to the content and displays information on the generator of the request.

3. The routing engine content receives information about the server. First, the routing engine content gets information about the server on the local device buffer via the adapter GET/PUT content, and when the device is local buffer does not contain the required content, the routing engine content additionally receives, through the name resolution mechanism, information on the node list containing the content needed by the user.

4. Collection module parameters receives information about the user's request from the generator of the query, gets the server information from the routing engine content, receives information on the state of the network (for example, information about background traffic and information communication lines) from the traffic generator, and obtains information about the output capacity and information about the input bandwidth peer subnet from the peer interface subnet.

5. The calculation engine uses the information received acquisition module settings, and a theory split optimization to calculate the joint optimization, optimal pair (user, server, and retrieves the route point-to-point generated by each pair (user, server, and speed of distribution of traffic for each route.

7. The routing engine content is best addressed from module output and converts it to change policy route and selection policy server, which can be realized and put into effect, where a policy route changes introduced in the transmission mechanism/redirect to perform routing policy on the lower level, and the server selection policy introduced in module support topology to perform server selection at the level of contents. In addition, the relevant information required to update the output bandwidth and input bandwidth, in addition to impose peer-to-peer network.

8. The mechanism of transfer/call forwarding adopts a policy change of a route from the routing engine content to modify the routing of the lower level and make policy changes to route IP level.

9. Support module topology adopts the policy server selection mechanism routing content to perform the selection of the application-tier server and establish the correspondence between the user and the server.

In this embodiment, the task WAN optimization equivalent transformed into several optimization problems subnet. In comparison with the joint task of optimization of the entire network, variables and constraints, the model of the joint optimization in the subnet are relatively small, and require a small amount of calculations. In this embodiment, opened the way for the model of joint optimization in the subnet. In accordance with the communication line, the server and the end point of the background traffic, the original mathematical model is broken down into smaller solved subproblems that reduces the execution time of the algorithm. Because subnet interact through bandwidth, the dependence of the interaction is relatively low, which is favourable for the solution of optimization problem.

The specialists in the art must understand that some or all of the stages of way according to the options of the implementation of the present invention can be implemented by software running on a suitable equipment. The program may be stored on computer media. When the program executes the stages of the process in accordance with the modalities for the implementation of the present invention. The media can be any media with the ability to store program code, such as ROM, RAM, floppy drive, or optical drive.

Finally, it should be noted that the above options for implementation provided only for the description of the technical solutions present invention, but do not limit the present invention. The specialists in the art must understand that although the present invention has been described in detail with reference to indicative options for implementation, can be made modifications and equivalent technical solutions present invention until then, until such modifications and replacement does not deviate from the ideas and technical solutions of the present invention.

1. The way for joint optimization of the network in which the network contains multiple subnets, and this method includes: obtaining information about the line of communications, information about the server and bandwidth, information about the user requirements in the subnet where information about the server and bandwidth contains throughput virtual server each external port subnet, and bandwidth virtual server is the bandwidth of the server on the subnet where the bandwidth of a server outside the subnet is required for the subnet via external port; for optimal routing and select parameter server subnet in accordance with the information about the line of communications, information about the server and capacity and information requirements; obtaining optimized input bandwidth each external port subnet in accordance with the option of optimal routing and select parameter server; compare optimized input bandwidth and throughput virtual server each external port, and if the results of the comparison between the optimized input bandwidth and throughput virtual server all external ports less than the threshold error, the application of parameter optimal routing and select parameter server on the subnet.

2. The method according to claim 1, wherein if all the results of the comparison between the optimized input bandwidth and throughput virtual server all external ports are not less than the threshold error, the process of obtaining optimized input bandwidth and comparing optimized input bandwidth and throughput virtual server to repeat until the results of the comparison will not be less than the threshold error.

3. The method according to claim 1, wherein the method of obtaining bandwidth virtual server external port contains: if you receive optimized output bandwidth neighboring subnet, the respective external port, throughput virtual server external port will be the least of optimized output capacity and available input bandwidth of the external port; if not received optimized output bandwidth neighboring subnet, the respective external port, throughput virtual server external port will be some value that is less than the available input bandwidth of the external port.

4. The method according to claim 1, further comprising: getting optimized bandwidth each external port subnet in accordance with a calculated parameter optimal routing and select parameter server subnet.

5. The method according to claim 1, wherein the receiving parameter optimal routing and select parameter server subnet in accordance with the information about the line of communications, information about the server and bandwidth, and user requirements includes: the installation of a mathematical model of joint optimization subnet; the execution of the split of first level for a mathematical model of joint optimization to obtain two subtasks SS-NBS and TE-NBS; performing a split second level two subtasks SS-NBS and TE-NBS in accordance with the communication line, the server and the end point of the background traffic, where split second level for subtasks SS-NBS provides the use of lines 1 and communication server's as structural elements to perform a split second level to obtain subtasks SS-NBL-L and SS-NBL-S, as well as the main tasks Dual SS-NBS , and the main task Dual SS-NBS manages subtasks SS-NBL-L and SS-NBL-S through the variables p l γ s and n t rates; dividing subtasks includes the use of communication lines 1 and end points j background traffic as structural elements to perform a split second level to obtain subtasks TE-NBS-L and TE-NBS-J, and the main task Dual TE-NBS , and the main task Dual TE-NBS manages subtasks TE-NBS-L and TE-NBS-J variables z l and t ij rates; and the calculation of the parameter optimal routing and select parameter server subnet in accordance with subjected to split mathematical model of joint optimization.

6. The method according to claim 1, wherein the optimised input bandwidth each external port subnet in accordance with the option of optimal routing and select parameter server contains: application parameter optimal routing and select parameter server subnet for calculation of user requirements for bandwidth virtual server via external port, and use a smaller value, the required bandwidth virtual server and available input bandwidth of the external port as optimized input throughput of the port subnet.

7. The method according to claim 4, which receive optimized output bandwidth each external port subnet in accordance with a calculated parameter optimal routing and select parameter server contains: application parameter optimal routing and select parameter server subnet to calculate the remaining bandwidth that can be provided by the server subnet via an external port to the outside world, and the use of lower values of the remaining bandwidth available output bandwidth of the external port as optimized output capacity of the port subnet.

8. The method according to claim 1 in which the information on the communication line contains at least one of the following: the bandwidth of the connection line, QoS lines of communication, the cost of the communication line.

9. The method according to claim 1, wherein information about the server and bandwidth contains: at least one of the following: a set of servers in the subnet and network bandwidth that can provide each server.

10. The method according to claim 1 in which information on the requirements of the user includes: at least one of the following: a set of users and a user requirement for the content.

11. A computer device joint optimization containing: a collection module settings made with the possibility of receiving information about the line of communications, information about the server and bandwidth requirements information of the user in the subnet where information about the server and bandwidth contains throughput virtual server each external port subnet, and bandwidth virtual server is the bandwidth of the server on the subnet where the bandwidth of a server outside the subnet is required for the subnet through external port; module of calculations performed with the option of optimal routing and select parameter server subnet in accordance with the information about the line of communications, information about the server and bandwidth, and information about the requirements of the user, and an optimized input bandwidth each external port subnet in accordance with the option of optimal routing and select parameter server; output module, made with the possibility of comparison optimized input bandwidth and throughput virtual server each external port, and if the results of the comparison between the optimized input bandwidth and throughput virtual server all external ports less than the threshold of an error, an application parameter optimal routing and select parameter server on the subnet, and the output parameter optimal routing and select parameter server.

12. The communications network, the network includes many subnets, and each subnet contains a system for joint optimization, in which the system for joint optimization gets information about a link, information about the server and bandwidth capacity, information on user requirements on a subnet with information about the server and bandwidth contains throughput virtual server each external port subnet; the parameter optimal routing and setting the server selection is calculated in accordance with the information about the line of communications, information about the server and bandwidth, information about the requirements of the user, and get optimized input bandwidth each external port subnet; and comparing optimized input bandwidth of each external ports and bandwidth virtual server external port, and if the comparison is less than the specified value error parameter optimal routing and selection option of the server used in your network.

13. Communications network indicated in paragraph 12, in which the subnet contains the subnet of the first level and one or more subnets top level, which subnet the first level contains at least one port to a higher-level network and the network of the upper level contains the port to the network infrastructure and the port to the network of higher level; since subnet first level calculate the parameter optimal routing and select parameter server subnet get optimized input bandwidth each external port subnet, and, finally, in high-level network calculates the optimal parameter routing and the option to select a server, and receive optimized input bandwidth each external port subnet; and comparing optimized input bandwidth of each external ports and bandwidth virtual server external port each subnet, and the result of the comparison is less than the specified value error parameter optimal routing and selection option of the server used in your network.

14. System for joint optimization containing: a computing device according to claim 11; and the routing of content, made with the possibility of conversion of routing option and select parameter server received computing device for the joint optimization of the routing option and the option to select the server used in the local network.