JP2016092652A - Asynchronous messaging server cooperation system evaluation device and evaluation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To evaluate an asynchronous messaging server cooperation system quickly and accurately at a low cost.SOLUTION: An asynchronous messaging server cooperation system of a network can be evaluated by accepting topology information and calculating a broker's load on the basis of the number of times that a shortest path tree from each publisher to all subscribers passes the broker, which is acquired from the accepted topology information (S3 and S4).SELECTED DRAWING: Figure 10

Description

  The present invention relates to an evaluation apparatus for an asynchronous messaging server cooperation method.

  As a method for realizing real-time data exchange, an asynchronous messaging technique called pub / sub is known. When pub / sub is used, it is generally composed of three parties: a publisher, a broker, and a subscriber.

  In such an asynchronous messaging method called pub / sub, load is easily concentrated on the broker. Therefore, load sharing is performed by linking a plurality of brokers.

  However, there are various methods of broker cooperation, and it is difficult to compare and evaluate these cooperation methods and select an appropriate method. Here, the evaluation is intended for the magnitude of the load on each broker and the uniformity of those loads. In other words, it can be said that the smaller and more uniform the load, the better the cooperation method.

  There are the following two methods for evaluating the conventional asynchronous messaging server linkage method.

1 Method of installing multiple physical machines and measuring the load status by actually linking the brokers (Non-Patent Documents 1 and 2)
2 Simple method to estimate the load on each broker (Non-patent Document 3)

“Performance Evaluation of Message-oriented Middleware using the SPECjms2007 Benchmark” (URL: http://www.cl.cam.ac.uk/research/srg/opera/publications/papers/SamKai08-PerfEvalJ-SPECjms2007.pdf) “Benchmarking Publish / Subscribe-based Messaging Systems” (URL: https://www.dvs.tu-darmstadt.de/publications/pdf/jms2009PS.pdf) “Logical topology construction method for routing to avoid congestion in data center” (URL: http://www.anarg.jp/achievements/web2013/paper/y-shimotsuma14IN-Configuration.pdf)

  However, in the method of installing multiple physical machines and actually measuring the load status by linking the broker, the accuracy of the evaluation is high, but the installation of the physical machines and the network connecting them costs a lot of cost. There is a problem that it takes a long time to appear. In addition, the said nonpatent literatures 1 and 2 are all about the evaluation about a single broker, and are not what evaluated about several brokers.

  In addition, the method of simply estimating the load on each broker can be realized at a low cost, but there is a problem that the accuracy of prediction is lowered because there is no established common scale used for estimation.

  The present invention has been made in view of the above circumstances, and provides an evaluation apparatus and an evaluation method for an asynchronous messaging server cooperation method capable of quickly and accurately evaluating an asynchronous messaging server cooperation method at low cost. With the goal.

  According to a first aspect of the present invention, a receiving unit that receives topology information indicating a connection relationship between the publisher, the broker, and the subscriber in a network having a publisher, a broker including an asynchronous messaging server, and a subscriber, and the receiving unit A network evaluation apparatus comprising load calculation means for calculating a load of each broker based on the number of times the shortest path tree from each publisher to all subscribers obtained from the received topology information passes through the broker. .

  According to a second aspect of the present invention, in the first aspect, the load of each broker is calculated separately for inbound and outbound.

  ADVANTAGE OF THE INVENTION According to this invention, the evaluation apparatus and evaluation method of an asynchronous messaging server cooperation system which can perform the evaluation of an asynchronous messaging server cooperation system with high accuracy quickly at low cost can be provided.

It is a figure for demonstrating the difference with a prior art and this invention. It is a figure which shows the structure of the network for demonstrating a mediation centrality. It is a diagram for explaining the calculation of Centrality of node v ₂ in the network shown in FIG. It is a figure for demonstrating the parameter used with the proposal parameter | index of this invention. It is a figure which shows an example of the structure of the network by which the evaluation by the evaluation apparatus which concerns on embodiment of this invention is performed. 6 is a diagram for explaining how to obtain PSG (b ₁ , b ₀ ) in the network configuration shown in FIG. 5. FIG. FIG. 6 is a diagram showing a relationship with the notation shown in FIG. 4 in the network configuration shown in FIG. 5. It is a figure which shows the functional block of the evaluation apparatus of the asynchronous messaging server cooperation system which concerns on embodiment of this invention. It is a figure for demonstrating the image which calculates | requires the load of each broker for every topology pattern. It is a flowchart for demonstrating operation | movement of the evaluation apparatus of the asynchronous messaging server cooperation system which concerns on this Embodiment.

<Summary of invention>
As described above, in the conventional technology, when associating a plurality of brokers in asynchronous messaging, it is costly to compare and evaluate the linkage methods according to actual machine measurement, and the method of simply estimating the load on each broker. However, there was a problem of lack of accuracy because there was no established common scale.

  The present invention extends the concept of mediation centrality to calculate the load for each broker in each linkage method, and makes it possible to estimate the load on each broker economically and with high accuracy. FIG. 1 is a diagram for explaining the difference between the prior art and the present invention, where (a) shows a method of actual machine measurement, (b) shows a simple estimation method, and (c) shows The technique of the present invention is shown.

  Next, “medium centrality” will be described.

“Medium centrality” is an index designed to quantify the influence of each node in a social network. “Medium centrality” takes the influence of each node as the number of times that the shortest path between nodes constituting the network passes. The intermediary centrality c _b (v _i ) of a node v _{i in} a network composed of n nodes is expressed by the following equation.

For example, mediated centrality of node v ₂ in the network shown in FIG. 2, as shown in FIG. 3, is calculated as 0.53. This value represents the influence of the node v _2.

  However, it is difficult to use the mediation centrality as it is as an evaluation index of an evaluation apparatus of the asynchronous messaging server cooperation method.

  This is because the role of nodes is not uniform in the case of pub / sub, unlike the network assumed for mediation centrality. Specifically, the node is divided into three types: publisher, broker, and subscriber, the centrality calculation target is only the broker, and the data flow direction is fixed in the order of publisher, broker, and subscriber. That is.

  In pub / sub, the relationship between publishers and subscribers is not one-to-one but many-to-many.

  Therefore, in the present invention, an extended index (hereinafter referred to as a proposed index) is newly defined as follows.

  (1) Consider the shortest path tree that is multicast from each publisher to all subscribers, and calculate the centrality based on the number of times that the shortest path tree passes through the broker.

(2) Since a tree is assumed, the centrality of each broker is inbound.
It is calculated by dividing into and outbound.

  In the following description, the notation shown in FIG. 4 is used.

Next, the PSG (b _i , b _j ) shown in FIG. 4 will be described.

Assume the network configuration of FIG. In the figure, the broker b ₀ is connected to the broker b ₁ , and the broker b ₁ is connected to the brokers b ₀ , b ₂ , and b ₃ . The broker b ₀ is connected to the publisher P ₀ and the subscriber S ₀ , the broker b ₁ is connected to the publisher P ₁ and the subscriber S ₁ , and the broker b ₂ is connected to the publisher P ₂ and the subscriber S _2. The publisher P ₃ is connected to the broker b ₃ .

For example, PSG (b ₁ , b ₀ ) represents “the number of publishers connected to the subgraph including b ₁ when the edge between b _{1 and} b ₀ is removed”. As shown in FIG. 6, the number of publishers included in the subgraph including b ₁ is “3”. Therefore, PSG (b ₁ , b ₀ ) = 3.

  FIG. 7 is a diagram showing a relationship with the notation shown in FIG. 4 in the network configuration shown in FIG.

  The present invention defines a new proposal index as follows.

For the inbound side (receiving side), an expression representing the load of the broker b _i is defined as follows.

For the outbound side (sender side), the formula representing the load on broker b _i is defined as follows:

The numerator of the expression representing the load on the broker b _i on the inbound side and the outbound side indicates the total number of times that the shortest path tree multicasted from each publisher to all subscribers passes through the broker.
<Embodiment>
Hereinafter, embodiments will be described with reference to the drawings.

  The evaluation apparatus of the asynchronous messaging server cooperation method according to the present embodiment is realized by an information processing apparatus that is a computer. The information processing apparatus has the same configuration as that of a normal computer including a memory, a CPU, an input unit, and a storage device, and the asynchronous messaging server cooperation method evaluation method according to the embodiment of the present invention is stored in the storage device. This is realized by the CPU executing an evaluation program that realizes the evaluation processing of the asynchronous messaging server cooperation method according to the embodiment of the present invention.

  FIG. 8 is a diagram showing functional blocks of the evaluation apparatus of the asynchronous messaging server cooperation method according to the embodiment of the present invention.

  As shown in FIG. 1, the information processing apparatus 1 that is an evaluation apparatus includes a topology candidate reception unit 11, a topology information generation and analysis unit 12, a load calculation unit 13, and a topology evaluation unit 14.

  The topology candidate receiving unit 11 receives at least one topology pattern that connects brokers.

  The topology information generation and analysis unit 12 generates topology information for each topology with respect to at least one topology pattern that connects the brokers received by the topology candidate reception unit 11.

  The topology information is connection information between brokers and connection information between brokers and publisher / subscribers. The connection information between brokers and the connection information between brokers and publishers / subscribers are represented by, for example, matrices. A mechanism for automatically generating such connection information from a topology diagram created by a GUI may be prepared. The topology information may be manually generated and input.

  FIG. 9 is a diagram for explaining an image for obtaining the load of each broker for each topology pattern. In the case of matrix representation, for example, the matrix G1 of the topology pattern T1 and the matrix G2 of the topology pattern T in FIG. It should be noted that the computer may be any form that is easy to handle, and is not necessarily in the form of a matrix.

Further, the topology information generation and analysis unit 12 uses parameters (P, S, NP (b _i ), NS () required by the load calculation unit (= calculation of the proposed index) based on the generated or input topology information. b _i ), B (b _i ), PSG (b _i , b _j ), IS (b _i , b _j )) are calculated.

  The load calculation unit 13 substitutes the parameters calculated by the topology information generation and analysis unit 12 into the formula of the proposal index, and calculates loads for each broker (inbound side load and outbound side load) for each topology. Note that the load calculation unit 13 may appropriately shape and output the calculated load for each broker.

  Thus, the topology can be evaluated by calculating the load (inbound side load and outbound side load) for each broker.

  The topology evaluation unit 14 evaluates the topology based on the load (inbound side load and outbound side load) for each broker calculated by the load calculation unit 13.

  In the evaluation of the topology, the more uniform the load of the broker, the more the load is not biased. Also, the smaller the value is 1.0 (which is the value of a single broker), the greater the load balancing effect. The topology evaluation unit 14 calculates the average value and standard deviation of the load of the broker for each topology. Then, superiority or inferiority is evaluated by comparing these values between topologies. The average value and the standard deviation may be calculated manually. Note that when there is one accepted topology pattern, there is no comparison target, and thus the comparative evaluation need not be performed.

  FIG. 10 is a flowchart for explaining the operation of the evaluation apparatus of the asynchronous messaging server cooperation method according to this embodiment.

  First, the topology candidate receiving unit 11 receives at least one topology pattern for connecting brokers (S1). Next, the topology information generation and analysis unit 12 generates topology information for each topology for at least one topology pattern that connects the brokers received by the topology candidate reception unit 11 (S2). Further, the processing of S1 and S2 may directly input the topology information regardless of the processing of the information processing apparatus.

Next, the topology information generation and analysis unit 12 uses parameters (P, S, NP (b _i ), NS required by the load calculation unit (= calculation of the proposed index) based on the generated or input topology information. (B _i ), B (b _i ), PSG (b _i , b _j ), IS (b _i , b _j )) are calculated (S3).

  Next, the load calculation unit 13 substitutes the parameters calculated by the topology information generation and analysis unit 12 into the formula of the proposal index, and calculates the load for each broker (inbound load and outbound load) for each topology. (S4). Thereby, since the load of each broker of the network for each topology can be calculated with high accuracy, the network topology can be evaluated with high accuracy.

  Next, the topology evaluation unit 14 evaluates the topology based on the load for each broker (inbound side load and outbound side load) calculated by the load calculation unit 13 (S5). The evaluation by the topology evaluation unit 14 is performed for each topology, for example, by calculating an average value and standard deviation of the broker load, and based on the calculated average value and standard deviation of the broker load. When there is a topology to be compared, the comparison is made by comparing the average value of the load of these brokers with the standard deviation.

  Therefore, according to the present embodiment, it is possible to calculate the load of the broker with high accuracy by using the proposed index. Therefore, it is possible to quickly evaluate the asynchronous messaging server cooperation method with high accuracy at low cost. it can.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  The method described in the embodiment is a program that can be executed by a computer (computer), for example, a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), an optical disk (CD-ROM, DVD, MO, etc.), It can be stored in a recording medium such as a semiconductor memory (ROM, RAM, flash memory, etc.), or transmitted and distributed via a communication medium. The program stored on the medium side includes a setting program that configures software means (including not only the execution program but also a table and data structure) in the computer. A computer that implements this apparatus reads a program recorded on a recording medium, constructs software means by a setting program as the case may be, and executes the above-described processing by controlling the operation by this software means. The recording medium referred to in this specification is not limited to distribution, but includes a storage medium such as a magnetic disk or a semiconductor memory provided in a computer or a device connected via a network.

v0-v6 nodes,
P1 to P3 Publisher S0 to S2 Subscriber 1 Information processing device 11 Topology information reception unit 12 Topology information generation and analysis unit 13 Load analysis unit 14 Topology evaluation unit

Claims (7)

Receiving means for receiving topology information indicating a connection relationship between the publisher, the broker, and the subscriber of a network having a publisher, a broker including an asynchronous messaging server, and a subscriber;
Network evaluation comprising load calculation means for calculating the load of each broker based on the number of times the shortest path tree from each publisher to all subscribers passes through the broker obtained from the topology information received by the reception means apparatus.   The network evaluation apparatus according to claim 1, wherein the load of each broker is calculated separately for inbound and outbound.   For each topology, calculate the average value and standard deviation of the calculated broker load, and compare the calculated average value and standard deviation of the broker load between topologies to evaluate the network topology. The network evaluation apparatus according to claim 1, further comprising topology evaluation means for performing. The network evaluation apparatus according to claim 2, wherein the load on the broker is calculated by the following equation.
Regarding the load on the inbound side,

For the outbound load,

here,
b _i: i-th broker P: Publisher Total S: total number of subscribers NP _{(b i):} b i number publisher is directly connected to the NS _{(b i):} b i number of subscribers B that are directly connected to (b _{i): b i} set PSG _{(b i} broker that is directly connected _{to, b j): b i -b} j number publisher connected to subgraph containing _{b i} in the case of removing the edges between iS _{(b i, b j):} the b i -b 1 if there are subscribers connected to the subgraph containing _{b i} in the case of removing the edges between _j, represents 0 if it does not exist. Accepting topology information indicating a connection relationship between the publisher, the broker and the subscriber of a network having a publisher, a broker including an asynchronous messaging server, and a subscriber;
A network evaluation comprising: a load calculation step of calculating a load of each broker based on the number of times that the shortest path tree from each publisher to all subscribers obtained from the topology information received by the reception step passes through the broker Evaluation method in the apparatus. In network evaluation equipment,
Accepting topology information indicating a connection relationship between the publisher, the broker and the subscriber of a network having a publisher, a broker including an asynchronous messaging server, and a subscriber;
A program for calculating the load of each broker based on the number of times the shortest path tree from each publisher to all subscribers obtained from the received topology information passes through the broker.   A computer-readable information recording medium storing the program according to claim 6.

Images