WO2014148247A1 - 処理制御システム、処理制御方法、および処理制御プログラム - Google Patents
処理制御システム、処理制御方法、および処理制御プログラム Download PDFInfo
- Publication number
- WO2014148247A1 WO2014148247A1 PCT/JP2014/055475 JP2014055475W WO2014148247A1 WO 2014148247 A1 WO2014148247 A1 WO 2014148247A1 JP 2014055475 W JP2014055475 W JP 2014055475W WO 2014148247 A1 WO2014148247 A1 WO 2014148247A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- processing
- analysis
- information
- time
- distribution
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/5044—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering hardware capabilities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/16—Error detection or correction of the data by redundancy in hardware
- G06F11/18—Error detection or correction of the data by redundancy in hardware using passive fault-masking of the redundant circuits
- G06F11/183—Error detection or correction of the data by redundancy in hardware using passive fault-masking of the redundant circuits by voting, the voting not being performed by the redundant components
- G06F11/184—Error detection or correction of the data by redundancy in hardware using passive fault-masking of the redundant circuits by voting, the voting not being performed by the redundant components where the redundant components implement processing functionality
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3409—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment
- G06F11/3419—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment by assessing time
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3409—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment
- G06F11/3433—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment for load management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/16—Error detection or correction of the data by redundancy in hardware
- G06F11/1675—Temporal synchronisation or re-synchronisation of redundant processing components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/503—Resource availability
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/508—Monitor
Definitions
- the present invention relates to a technique for distributing processing to a plurality of computers.
- CEP Complex Event Processing
- batch processing has a wide range of data to be processed (from several days to several years).
- the CEP technique targets the latest data range (a few minutes). Due to the difference in properties, batch processing is suitable for trend analysis and the like, and CEP technology is suitable for illegal value detection and real-time billing control based on access logs and the like.
- the analysis server analyzes a large amount of data in a CEP technology system, the processing performance of the analysis server is limited.
- the amount of data processed by the CEP technology is particularly large, there is a risk that performance degradation may occur due to a delay in data analysis processing by the CEP technology.
- Patent Document 1 paragraphs 0039 and 0041.
- Patent Document 1 is a mechanism for selecting an analysis server with a low processing load and executing the process in order to prevent performance degradation.
- Patent Document 2 a technique for distributing the load among a plurality of computers has been considered (Patent Document 2).
- a plurality of computers express each load as a queue usage rate, and attempt to prevent load bias by transferring a client request to a computer having the lowest usage rate.
- redundancy is one of the mechanisms for improving reliability, that is, providing fault tolerance (fault tolerance).
- fault tolerance fault tolerance
- a plurality of the same systems are prepared, and all of them are executed in parallel, and even if a failure occurs in a certain system, reliability is ensured by continuing operation of other systems.
- the degree of failure for example, the occurrence of a calculation error due to an internal failure of the CPU
- the result may not be the same even if the same processing is performed in a plurality of systems.
- Patent Document 3 in a system that receives a processing request and transmits the result of processing the request by a computer, the same processing request is transmitted to all of the plurality of computers, and the processing result is received from the plurality of computers.
- a configuration for comparing a plurality of reception results and outputting a correct processing result is shown.
- Patent Document 1 aggregates highly related events to a specific analysis server based on the load. Therefore, when analysis target data is particularly large, the load on the analysis server may limit the processing performance. It becomes difficult to complete the analysis within a predetermined time.
- Patent Document 2 the technology for transferring requests between a plurality of computers disclosed in Patent Document 2 is difficult to perform load distribution efficiently. Further, when processing requests are processed between a plurality of computers having different specifications as described above, a difference in load balance occurs between the plurality of computers, so that a time difference in request processing occurs.
- Patent Documents 1 and 2 only select one server or computer with a low processing load, and do not consider fault occurrence, so they do not have fault tolerance.
- the processing is executed in parallel on all of a plurality of computers.
- the upper limit of the processing performance of one computer determines the performance of the entire system. Therefore, when the technique disclosed in Patent Document 3 is applied to a system that analyzes and analyzes a large amount of information such as big data, there is a problem that the throughput of the system decreases as the processing amount increases.
- n computers for example, an analysis server
- m Provides a technology that selects a computer (for example, an analysis server) (1 ⁇ m ⁇ n) and processes a plurality of processing results by the selected computer (for example, an analysis server) by a predetermined method to obtain the most probable result. Is done.
- a processing system that is a more specific aspect includes the following configuration.
- a load distribution device connected to n (n is 3 or more) processing devices, a processing result determination device connected to the n processing devices, and the load distribution device and the processing result determination device A management device.
- the management device acquires time information required in the past processing for each of the n processing devices, and determines the time required for each of the n processing devices to process the processing target information. And estimating time information indicating the relationship between the estimated time and the processing device is transmitted to the load distribution device.
- the load distribution apparatus determines m (1 ⁇ m ⁇ n) processing apparatuses that process the processing target information based on the estimated time information, and determines the processing target information as the determined m pieces of processing target information. Send to the specified processing device.
- Each of the m designated processing devices calculates a processing result by processing the processing target information, and transmits the processing result to the processing result determination device.
- the processing result determination device receives m processing results from the m designated processing devices, estimates a true value from the m processing results by a predetermined method, and determines the estimated value. The estimated value is transmitted to a subsequent processing apparatus that processes
- a load distribution program for a computer of a load distribution apparatus connected to n which is another aspect, is the management apparatus connected to the load distribution apparatus, For each of the n processing devices, time information required in the past processing is acquired, and a time required for processing the processing target information by each of the n processing devices is estimated based on the time information.
- the estimated time information indicating the relationship between the estimated time and the processing device is transmitted to the load balancer, m pieces (1 ⁇ m ⁇ n) for processing the processing target information based on the estimated time information.
- Is executed by the computer and the computer is caused to transmit the processing target information to the determined m designated processing devices.
- Each of the m designated processing devices calculates a processing result by processing the processing target information, and transmits the processing result to the processing result determination device.
- the processing result determination device receives m processing results from the m designated processing devices, estimates a true value from the m processing results by a predetermined method, and determines the estimated value. The estimated value is transmitted to a subsequent processing apparatus that processes
- a processing result determination program for a computer of a processing result determination apparatus connected to n which is another aspect, is a management apparatus connected to the processing result determination apparatus
- time information required in the past processing is acquired, and the required time required for processing of the processing target information by each of the n processing devices is determined based on the time information.
- Estimated time is estimated, and estimated time information indicating the relationship between the estimated time and the processing device is transmitted to the management device, the n processing devices, and the load distribution device connected to the processing result determination device.
- the load balancer determines m (1 ⁇ m ⁇ n) processing devices for processing the processing target information based on the estimated time information, and determines the m processing target information. Specified place When each of the m designated processing devices calculates the processing result by processing the processing target information and transmits the processing result to the processing result determination device, the m designation processing devices Causing the computer to receive m processing results from the processing device, respectively, and causing the computer to estimate a true value from the m processing results by a predetermined method to obtain an estimated value; The computer is caused to transmit the estimated value to a subsequent processing apparatus that processes the estimated value.
- processing can be distributed so that processing results are output within the required time.
- the administrator weights the processing results of a certain server, and a majority decision based on the weighting is performed. You can also get true results.
- the processing results of a server are invalidated. As a result, it is possible to obtain a true result by majority decision based on this.
- a probability distribution or an average value when the result is output in multiple values is used as a mechanism for estimating a true result of processing results in a plurality of servers. It is possible to obtain a probable estimated value by providing a mechanism for obtaining an estimated value and a mechanism for determining even if the number of results for determination is an even number.
- 1 shows a configuration of a computer system used in each embodiment.
- 1 shows a configuration of a dispersion apparatus 10.
- the structure of the management server 9 is shown.
- the structure of the analysis server 11 is shown.
- the structure of the log output server 8 is shown.
- the structure of the determination apparatus 12 is shown.
- the distribution management table 121 in the distribution apparatus 10 is shown.
- An event type identification table 122 in the distribution apparatus 10 is shown.
- 2 shows a communication management table 123 in the distribution apparatus 10.
- a request time table 920 in the management server 9 is shown.
- An analysis logic management table 921 in the management server 9 is shown.
- the connection destination management table 922 in the management server 9 is shown.
- the time management table 923 in the management server 9 is shown.
- the statistical time management table 924 in the management server 9 is shown.
- the 2 shows a logic management table 1120 in the analysis server 11.
- the queue management table 1121 in the analysis server 11 is shown.
- the connection destination management table 1123 in the analysis server 11 is shown.
- a simple distribution rule 820 in the log output server 8 is shown.
- a simplified distribution destination table 821 in the log output server 8 is shown.
- the analysis result management table 720 in the determination apparatus 12 is shown.
- the distribution number management table 721 in the management server 9 and the determination apparatus 12 is shown.
- 7 shows a communication management table 722 in the determination apparatus 12.
- the distribution number information transmission processing by the management server 9 is shown.
- the queue statistical information transmission process by the management server 9 is shown.
- the distribution order update process by the distribution apparatus 10 is shown.
- the event transmission process by the log output server 8 is shown.
- the distributed processing by the distribution apparatus 10 is shown.
- the distribution destination response process by the management server 9 is shown.
- the distribution destination registration process by the management server 9 is shown.
- the analysis process by the analysis server 11 is shown.
- the distribution number information update process by the determination apparatus 12 is shown.
- the analysis result determination process by the determination apparatus 12 is shown.
- the analysis result registration process by the management server 9 is shown.
- the structure of the management server 9 of Example 2 is shown. 9 shows an analysis server state management table 925 in the management server 9 of the second embodiment.
- the structure of the dispersion apparatus 10 of Example 3 is shown.
- 10 shows a session management table 124 in the distributed apparatus 10 of the third embodiment.
- distribution apparatus 10 of Example 3 is shown.
- Each of the n processing devices corresponds to an analysis device, an analysis server 11, and the like. Processing corresponds to analysis and the like.
- the load distribution device corresponds to the distribution device 10 or the like.
- the processing result determination device corresponds to the analysis result determination device, the determination device 12, and the like.
- the management device corresponds to the management server 9 or the like.
- the subsequent processing device corresponds to the processing device that processes the estimated value, the AP server 13, and the like.
- the processing target information corresponds to analysis target information or an event.
- Each of the m designated processing devices corresponds to a specific analysis device, a distribution destination analysis server 11, and the like.
- the time information corresponds to the time management table 923, estimated time, and the like.
- the estimated time corresponds to the statistical time management table 924, estimated time, average estimated time, and the like.
- the estimated time information corresponds to queue statistical information, distributed destination responses, and the like.
- the load distribution program corresponds to the program of the distribution apparatus 10 and the like.
- the processing result determination program corresponds to the program of the determination device 12 or the like.
- the communication device corresponds to the log output server 8 or the like.
- the processing control program corresponds to a program of the management server 9 or the like.
- the upper limit corresponds to the request time.
- FIG. 1 shows the configuration of a computer system according to an embodiment of the present invention.
- the computer system includes a plurality of clients 7, a plurality of log output servers 8, a management server 9, a plurality of distribution devices 10, a plurality of analysis servers 11, a plurality of determination devices 12, and a plurality of determination devices. And an AP (Application) server 13.
- the number of log output servers 8 may be one, the number of distribution devices 10 may be one, the number of determination devices 12 may be one, or an AP server The number of 13 may be one.
- the client 7 may be omitted.
- the client 7 is connected to the log output server 8 via the network 15.
- the log output server 8 is connected to a plurality of distributed devices 10 via the network 16.
- the distribution apparatus 10 is connected to a plurality of analysis servers 11 via the network 17.
- the analysis server 11 is connected to the plurality of determination devices 12 via the network 18.
- the determination device 12 is connected to a plurality of AP servers 13 via the network 19.
- the management server 9 is connected to a plurality of distributed devices 10 via the network 5.
- the management server 9 is connected to a plurality of analysis servers 11 via the network 6.
- the management server 9 is connected to a plurality of determination devices 12 via the network 14. Note that the management server 9 may not be connected to the plurality of analysis servers 11.
- the distribution device 10 and the determination device 12 are provided as separate hardware. However, the distribution device 10 and the determination device 12 may be provided in the same hardware. . Further, the distribution device 10, the analysis server 11, and the determination device 12 may be provided in the same hardware.
- the management server 9 is provided as hardware different from the distribution device 10, the analysis server 11, and the determination device 12. However, the management server 9 includes the distribution device 10, the analysis server, and the like. 11 or the determination device 12 may be provided.
- the management server 9 is provided as a single computer. However, the management server 9 may be provided as a computer system having a plurality of management servers 9. However, in that case, it is necessary to share or synchronize information among the plurality of management servers 9.
- the client 7 communicates with the log output server 8.
- the log output server 8 generates a log indicating the content of communication with the client 7, selects a transmission destination from among a plurality of distributed devices 10 according to the content of the event in the log, and sends it to the selected distributed device 10. Output an event.
- the distribution apparatus 10 that has received the event selects a transmission destination from the plurality of analysis servers 11 according to the contents of the event, and transmits the event to the selected analysis server 11.
- the analysis server 11 that has received the event analyzes the event and transmits analysis result information including an analysis value obtained from the analysis to the determination device 12.
- the determination device 12 that has received the plurality of analysis values estimates the true value of the plurality of analysis values as an estimation value, and transmits the estimation value as a final analysis result to the AP server 13 that requests the analysis result. .
- the AP server 13 that has received the estimated value executes an AP that uses the estimated value.
- the client 7 corresponds to a communication terminal device such as a mobile phone or a smartphone used by the user.
- the log output server 8 corresponds to a base station, a GW (Gateway), and the like, and outputs a communication log of the communication terminal device as an event.
- the analysis server 11 analyzes the communication amount for each user, the usage amount of paid content, and the like based on the event.
- the AP in the AP server 13 controls the bandwidth of the communication terminal device according to the communication amount of each communication terminal device and the billing process for processing the billing to the user according to the usage of the paid content and the communication amount of the communication terminal device. Perform bandwidth control and so on.
- the log output server 8 may be a camera or a sensor. In this case, the log output server 8 transmits the detected information to the distribution apparatus 10 as an event.
- the computer system in this case may not have the client 7.
- Such a computer system can also be applied to a network system that communicates data between a server and a client, such as a WWW (World Wide Web), a mail system, and a data center.
- a WWW World Wide Web
- a mail system such as a mail system, and a data center.
- a system including the management server 9, the distribution device 10, and the determination device 12 may be referred to as an analysis control system.
- a system having an analysis control system and a plurality of analysis servers 11 may be referred to as an analysis system.
- FIG. 2 shows the configuration of the dispersion apparatus 10.
- the distributed device 10 is realized by a computer having one or more CPUs (Central Processing Unit) 101, one or more network interfaces (NW I / F) 102 to 104, an input / output device 105, and a memory 107. . Each part of the distribution device 10 is connected to each other via a communication path such as the internal bus 106.
- the NW I / F 102 is connected to the log output server 8 via the network 16.
- the NW I / F 103 is connected to the analysis server 11 via the network 17.
- the NW I / F 104 is connected to the management server 9 via the network 6.
- the memory 107 stores programs of the distributed function 111 and the logic load acquisition function 112, a distributed management table 121, an event type identification table 122, and a communication management table 123.
- Each program of the distribution apparatus 10 may be stored in the memory 107 of the distribution apparatus 10 in advance, or may be recorded on a recording medium that can be used by the distribution apparatus 10 and introduced into the memory 107 when necessary.
- the recording medium refers to a storage medium that can be attached to and detached from an external device interface (not shown) in the distribution apparatus 10, for example.
- the distribution function 111 determines the analysis server 11 as a distribution destination of the event received from the log output server 8 based on the distribution management table 121, and transfers the event to the distribution destination.
- FIG. 3 shows the configuration of the management server 9.
- the management server 9 is realized by a computer having one or more CPUs 901, one or more network interfaces (NW I / F) 902 to 904, an input / output device 905, and a memory 907. Each unit of the management server 9 is connected to each other via a communication path such as the internal bus 906.
- the NW I / F 902 is connected to the distribution apparatus 10 via the network 5.
- the NW I / F 903 is connected to the determination device 12 via the network 14.
- the NW I / F 904 is connected to the analysis server 11 via the network 6.
- the memory 907 includes programs for the management function 910 and logic management function 911, a request time table 920, an analysis logic management table 921, a distributed number management table 721, a connection destination management table 922, a time management table 923, and a statistical time management table. 924 are stored.
- each program of the management server 9 may be stored in advance in the memory 907 of the management server 9, or recorded in a recording medium that can be used by the management server 9, and when necessary, the memory 907. May be introduced.
- the logic management function 911 acquires information related to distribution from the distribution device 10 and the determination device 12, and analyzes the queue state of the analysis server 11 based on the information. Then, the management function 910 determines a distribution destination according to the request time of the AP based on the result analyzed by the logic management function 911. The management function 910 receives an instruction from a computer system administrator.
- FIG. 4 shows the configuration of the analysis server 11.
- the analysis server 11 is realized by a computer having one or more CPUs 1101, one or more network interfaces (NW I / F) 1102 to 1104, an input / output device 1105, and a memory 1107. Each part of the analysis server is connected to each other via a communication path such as the internal bus 1106.
- the NW I / F 1102 is connected to the distribution apparatus 10 via the network 17.
- the NW I / F 1103 is connected to the determination device 12 via the network 18.
- the NW I / F 1104 is connected to the management server 9 via the network 6.
- the memory 1107 stores the analysis function 1110 and queue management function 1111 programs, a logic management table 1120, a queue management table 1121, an event type identification table 122, and a connection destination management table 1123.
- each program of the analysis server 11 may be stored in advance in the memory 1107 of the analysis server 11, or is recorded on a recording medium that can be used by the analysis server 11, and when necessary, the memory 1107 is stored. May be introduced.
- the analysis function 1110 selects an analysis logic for analyzing an event received from the distributed device 10 based on the event type identification table 122 and the logic management table 1120, and executes an analysis process according to the analysis logic.
- the queue management function 1111 once stores the event received from the distribution apparatus 10 in the queue management table 1121 and extracts the event from the queue management table 1121 as required by the analysis function 1110.
- the queue management table 1121 in the analysis server 11 may be simply referred to as a queue.
- FIG. 5 shows the configuration of the log output server 8.
- the log output server 8 is realized by a computer having one or more CPUs 801, one or more network interfaces (NW I / F) 802 to 803, an input / output device 804, and a memory 806. Each unit of the log output server 8 is connected to each other via a communication path such as the internal bus 805.
- the NW I / F 802 is connected to the client 7 via the network 15.
- the NW I / F 803 is connected to the distribution apparatus 10 via the network 16.
- the memory 806 stores each program of the node function 810 and the simple distribution function 811, a simple distribution rule 820, a simple distribution destination table 821, and an event type identification table 122.
- each program of the log output server 8 may be stored in advance in the memory 806 of the log output server 8, or recorded on a recording medium that can be used by the log output server 8, and when necessary. May be installed in the memory 806.
- the node function 810 executes processing for the request received from the client 7.
- the simple distribution function 811 transfers the access log output in the process in which the request of the client 7 is processed by the node function 810 to the distribution apparatus 10 from among the distribution apparatuses 10 registered in the simple distribution destination table 821. Then, the distribution device 10 is selected and distributed according to the simple distribution rule 820.
- FIG. 6 shows the configuration of the determination device 12.
- the determination device 12 is realized by a computer having one or more CPUs 1201, one or more network interfaces (NW I / F) 1202 to 1203, an input / output device 1204, and a memory 1206.
- the determination devices 12 are connected to each other via a communication path such as the internal bus 1205.
- the NW I / F 1202 is connected to the analysis server 11 via the network 18.
- the NW I / F 1203 is connected to the AP server 13 via the network 19.
- the memory 1206 stores a program for the determination function 1210, an analysis result management table 720, a distribution number management table 721, a request time table 920, and a communication management table 722.
- each program may be stored in advance in the memory 1206 of the determination device 12, or recorded in a recording medium that can be used by the determination device 12 and introduced into the memory 1206 when necessary. Also good.
- the determination function 1210 stores the analysis result information received from the plurality of analysis servers 11 in the analysis result management table 720, and estimates the analysis value with the highest probability based on the analysis values indicated in the plurality of analysis result information. And the estimated value is transmitted to the AP server 13.
- FIG. 7 shows a distribution management table 121 in the distribution apparatus 10.
- the distribution management table 121 is information for distributing an event received from the log output server 8 by the distribution apparatus 10, and indicates a distribution instruction according to the event classification.
- the distributed management table 121 has a record for each event type.
- the event type indicates a classification of request time.
- the request time is an upper limit of processing time (required time) that can be used for analysis, and is requested by the AP.
- the event types in this embodiment are a fixed time type and a real time type.
- the request time of the real time type is shorter than the request time of the fixed time type.
- Each record of the distribution management table 121 has an event type column 1211, a distribution order column 1212, an update time column 1213, and a distribution number column 1214.
- the event type is registered in the event type column 1211.
- the contents registered here are the contents of the event type column 1221 of the event type identification table 122 described later.
- the distribution order column 1212 priorities of a plurality of analysis servers 11 that are candidates for distribution of the event type event are registered.
- the update time column 1213 the update time of the record is registered.
- the distribution number column 1214 a distribution number that is the number of analysis servers 11 that are the distribution destinations of the event type event is registered.
- the distribution apparatus 10 when the event type is “fixed time type”, the distribution apparatus 10 is distributed to “5” analysis servers 11 that are the number of distributions.
- the distribution apparatus 10 selects as many analysis servers 11 as the distribution destination in order from the top registered in the distribution order column 1212. Note that the number of distributions may be smaller than the number of analysis servers 11 registered in the distribution order column 1212.
- FIG. 8 shows the event type identification table 122 in the distributed apparatus 10.
- the event type identification table 122 is information for identifying an event received from the log output server 8.
- the event type identification table 122 has a record for each event type. Each record has an event type column 1221, a protocol column 1222, and a header column 1223.
- the event type discriminated in the computer system is registered in the event type column 1221.
- the protocol column 1222 the protocol name of the event received from the log output server 8 is registered.
- the header field 1223 the name of a specific header referred to for identifying the event type is registered.
- the distribution apparatus 10 identifies the corresponding event type by referring to the event protocol and a specific header.
- FIG. 9 shows the communication management table 123 in the distributed apparatus 10.
- the communication management table 123 IP addresses and netmasks necessary for the distributed device 10 to communicate with the connection destination are registered.
- the connection destination is the analysis server 11, the management server 9, or the like.
- the communication management table 123 has a record for each connection destination. Each record has a distinction column 1231, an address column 1232, and a mask value column 1233.
- connection destination In the distinction column 1231, the name of the connection destination for identifying the connection destination is registered. In the address column 1232 and the mask value column 1233, the IP address and netmask value of the connection destination are registered.
- FIG. 10 shows a request time table 920 in the management server 9.
- the request time table 920 a request time that is an analysis time required by the AP is registered.
- the request time table 920 has a record for each event type. Each record has an event type column 9201 and a request time column 9202.
- the event type is registered in the event type column 9201.
- the request time column 9202 a request time for the event type is registered.
- the request of the AP server 13 is to request that each estimated value be transferred to the AP server 13 within the required time “500 ms” from the occurrence of the event.
- three or more event types may be registered in the request time table 920 corresponding to three or more request times, respectively.
- FIG. 11 shows an analysis logic management table 921 in the management server 9.
- the analysis logic management table 921 is information related to the analysis logic executed by each analysis server 11. Based on this information, the management server 9 calculates how long it takes the analysis processing of one event by the analysis server 11.
- the analysis logic management table 921 has a record for each combination of event type, analysis server, and analysis logic. Each record includes an event type column 9211, an analysis server column 9212, an analysis server 11, and an analysis logic column 9213.
- the event type is registered in the event type column 9211.
- a name for identifying the analysis server 11 is registered.
- analysis logic for the event type of the analysis server 11 is registered.
- analysis time column 9214 an analysis time that is a time required for the analysis processing when the analysis server 11 executes the analysis logic is registered.
- the analysis time required for the analysis server 11 of the “analysis server a” to process the real-time event by the analysis logic “R1” is “123 ms”.
- the management server 9 receives the information indicating the distribution of the event by the distribution device 1 from the distribution device 10 when the queue of the analysis server 11 is empty, and the result of the analysis of the event by the determination device 12. The difference from the time at which the information indicating the reception is received from the determination device 12 is calculated as the analysis time.
- the time required may be calculated as the analysis time by the distributed device 10, the management server 9, or the like transmitting a pseudo event to the analysis server 11 and acquiring the analysis result.
- the performance of each of the plurality of analysis servers 11 may be different as described above.
- the time required for the event analysis process differs. That is, the time required for the event analysis process includes the performance information of the analysis server 11.
- FIG. 12 shows a connection destination management table 922 in the management server 9.
- the connection destination management table 922 contains information on IP addresses and netmasks necessary for the management server 9 to communicate with the connection destination.
- the connection destination is the distributed device 10, the determination device 12, the analysis server 11, and the like.
- the connection destination management table 922 has a record for each connection destination. Each record has a distinction column 9221, an address column 9222, and a mask value column 9223.
- the name of the connection destination is registered.
- the address column 9222 and the mask value column 9223 the IP address and netmask value of the connection destination are registered.
- FIG. 13 shows a time management table 923 in the management server 9.
- the time management table 923 is information for grasping how many events are registered in the queue of the analysis server 11. Further, the time management table 923 indicates estimated time used for determining a real-time type distribution destination.
- the management server 9 includes information indicating to which analysis server 11 the event is distributed from the distribution device 10 and the determination device 12 that are before and after the analysis server 11, and information indicating from which analysis server 11 the analysis result information is obtained. Are received and sequentially registered in the time management table 923.
- the management server 9 can estimate the queue status of each analysis server 11 by referring to the time management table 923.
- the time management table 923 includes a record for each information received from the distribution device 10 or the determination device 12 by the management server 9. Each record has an analysis server column 9231, a reception source column 9232, an event type column 9233, an event type column 9234, an estimated time column 9235, and an update time column 9235.
- the estimated time corresponds to the estimated time.
- the identification information of the analysis server 11 indicated in the information received from the distribution apparatus 10 or the identification information of the analysis server 11 indicated in the information received from the determination apparatus 12 is registered.
- the receiving source column 9232 information indicating whether the information of the record is received from the distribution device 10 or the determination device 12 is registered.
- the event type column 9233 the event type of the event indicated in the information received from the distributed apparatus 10 or the event type of the event indicated in the information received from the distributed apparatus 10 is registered.
- the event type column 9234 the event type of the event is registered.
- the event type is identification information of the client 7, identification information of the user of the client 7, and the like.
- the estimated time column 9235 an estimated time from when an event is put into the queue to when the analysis is completed when processing is executed with the analysis logic corresponding to the event type of the record in the analysis logic management table 921 is registered.
- the In the estimated time column 9235 the estimated time received from the determination device 12 is not registered.
- the update time column 9235 the time when the record is updated is registered.
- the management server 9 updates the time management table 923 by an analysis result registration process described later.
- the event type is identification information of a communication terminal device.
- the analysis server 11 analyzes the amount of communication according to the event type, it is possible to determine charging or bandwidth control for a specific communication terminal device.
- the second record indicates that the distribution device 10 transmits the event of “real time type” and the identifier “B” to the “analysis server b”, and then the determination device 12 receives the analysis result information from the “analysis server b”.
- the estimated time until reception is “0.123 sec”.
- the fourth record 0 indicates that the determination device 12 has received the analysis result information of the event indicated in the second record from the “analysis server b”.
- the management server 9 can estimate the state of each analysis server 11 by scanning the time management table 923.
- FIG. 14 shows a statistical time management table 924 in the management server 9.
- the statistical time management table 924 indicates an average estimated time used for determining a fixed time type distribution destination.
- the statistical time management table 924 includes a record for each combination of the analysis server 11 and the analysis logic. Each record has an analysis server column 9241, an average estimated time column 9242, an update time column 9243, a total time column 9244, a counter column 9245, and an analysis logic 9246.
- identification information of the analysis server 11 is registered.
- the average estimated time column 9242 the average estimated time from when the event of the analysis logic is newly input to the current analysis server 11 until the analysis result information is output is the total time column 9244, the counter column 9245, Is calculated and registered based on the information. Note that the average estimated time may not include the analysis time of the newly input event.
- the update time column 9243 the time when the record was updated is registered.
- the total time column 9244 registers the total time, which is a cumulative value of the time from when an event is input to the analysis logic of the analysis server 11 until the analysis is completed.
- the counter column 9245 the counter value of the event input to the analysis logic of the analysis server 11 is registered.
- the total time and the counter value may be a cumulative value from the start of the operation of the analysis server 11 or may be periodically reset.
- identification information of the analysis logic is registered. Corresponding event type identification information may be registered instead of the analysis logic.
- the management server 9 updates the statistical time management table 924 by an analysis result registration process described later.
- the average estimated time registered in the average estimated time column 9242 is a quotient obtained by dividing the total time in the total time column 9244 by the counter value in the counter column 9245. That is, the average estimated time is an average value of the time from when an event of the analysis logic of the analysis server 11 is input until the analysis result information is output. In this way, in the statistical time management table 924, a statistical average estimated time when an event is processed in each analysis server 11 is registered. By using the statistical time management table 924, the management server 9 can perform distribution satisfying the requested time without grasping the queue status of the analysis server 11 in units of events.
- the statistical time management table 924 may have a record for each analysis server 11.
- FIG. 15 shows a logic management table 1120 in the analysis server 11.
- the logic management table 1120 indicates analysis logic for each event received from the distribution apparatus 10 by the analysis server 11.
- the logic management table 1120 has a record for each event type. Each record has an event type column 11201 and an analysis logic column 11203.
- event type column 11201 an event type is registered.
- analysis logic column 11203 an analysis logic for analyzing the event type event is registered.
- FIG. 16 shows the queue management table 1121 in the analysis server 11.
- the analysis server 11 sequentially registers the events received from the distribution apparatus 10 in the queue management table 1121. Further, the analysis server 11 sequentially executes events from the top of the queue management table 1121.
- the queue management table 1121 has a record for each event. Each record has a number (No.) column 11211, an analysis logic column 11212, an event content column 11213, and a reception time column 11214.
- the serial number of the event is registered.
- the analysis logic column 11212 the analysis logic of the event is registered.
- the event content column 11213 the event content received from the distribution apparatus 10 is registered.
- the event content is all the data of the received event.
- the event content is received binary data.
- the reception time column 11214 the time when the record is registered is registered.
- FIG. 17 shows the connection destination management table 1123 in the analysis server 11.
- the connection destination management table 1123 is information on the IP address and netmask necessary for the analysis server 11 to communicate with the connection destination.
- the connection destination is the distributed device 10, the determination device 12, the management server 9, and the like.
- the connection destination management table 1123 has a record for each connection destination. Each record has a distinction column 11231, an address column 11232, and a mask value column 11233.
- the name of the connection destination is registered.
- the address column 11232 and the mask value column 11233 the IP address and netmask value of the connection destination are registered.
- FIG. 18 shows a simple distribution rule 820 in the log output server 8.
- the log output server 8 selects an event transmission destination from a plurality of distributed devices 10 in accordance with the simple distribution rule 820.
- the simple distribution rule 820 has a rule column 8201. In the rule column 8201, distributed logic for selecting an event transmission destination is registered.
- FIG. 19 shows a simple distribution destination table 821 in the log output server 8.
- the simple distribution destination table 821 IP address and netmask information necessary for the log output server 8 to transmit an event to the distribution apparatus 10 are registered.
- the simple distribution destination table 821 has a record for each distribution apparatus 10. Each record has a distinction column 8211, an address column 8212, and a mask value column 8213.
- the name of the distribution apparatus 10 as the transmission destination is registered.
- the address column 8212 and the mask value column 8213 the IP address and netmask value of the registration destination name of the record are registered.
- FIG. 20 shows an analysis result management table 720 in the determination apparatus 12.
- the determination device 12 registers the analysis result information received from the analysis server 11 in the analysis result management table 720, determines a majority decision based on the analysis result management table 720, determines an estimated value, and estimates the estimated value. Is transmitted to the AP server 13.
- the analysis result management table 720 has a record for each analysis result information. Each record has a number (No.) column 12201, an event type column 12202, an analysis server column 12203, an event type column 12204, an analysis value column 12205, and a reception time column 12206.
- the serial number of the analysis result information is registered in the number column 12201.
- the event type column 12202 the event type of the target event of the analysis result information is registered.
- the analysis server column 12203 the identification information of the analysis server 11 that has output the analysis result information (the transmission source of the analysis result information) is registered.
- the event type column 12204 the event type of the event is registered.
- the analysis value column 12205 an analysis value that is a result of the analysis indicated in the analysis result information and output by the analysis by the analysis server 11 is registered.
- the reception time column 12206 the time when the analysis result information is received is registered.
- the determination device 12 identifies each event using two pieces of information in the analysis server column 12203 and the event type column 12204, and performs determination using the identified event.
- FIG. 21 shows a distribution number management table 721 in the management server 9 and the determination device 12.
- the distributed number management table 721 indicates the distributed number for each event type.
- the determination device 12 determines the timing for performing the majority decision on the analysis result information based on the distribution number management table 721 and the number of analysis result information.
- the distributed number management table 721 has a record for each event type. Each record has an event type column 12211 and a distribution number column 12212.
- the event type is registered in the event type column 12211.
- the distribution number column 12212 a distribution number that is the number of analysis servers 11 that are the distribution destinations of the event type event is registered.
- FIG. 22 shows a communication management table 722 in the determination apparatus 12.
- the communication management table 722 is information on IP addresses and netmasks necessary for the determination apparatus 12 to communicate with the connection destination.
- the connection destination is the management server 9, the AP server 13, or the like.
- the communication management table 722 has a record for each connection destination. Each record has a distinction column 12231, an address column 12232, and a mask value column 12233.
- the name of the connection destination is registered.
- the IP address and netmask value of the connection destination are registered.
- FIG. 23 shows distribution number information transmission processing by the management server 9.
- the management function 910 of the management server 9 registers the contents of the distributed number management table 721 according to the input from the administrator.
- the management function 910 performs distribution number information transmission processing for transmitting the contents of the distribution number management table 721 to the distribution device 10 and the determination device 12.
- the management function 910 acquires the contents of the distribution number management table 721 (step S341). Next, the management function 910 transmits the information acquired in step S341 as distribution number information to the distribution device 10 and the determination device 12 (step S342).
- the distribution number information indicates the distribution number for each event type. The above is the distributed number information transmission processing.
- FIG. 24 shows queue statistical information transmission processing by the management server 9.
- the management function 910 of the management server 9 performs queue statistical information transmission processing for periodically transmitting queue statistical information indicating the queue status of the analysis server 11 to the distribution device 10 based on the statistical time management table 924.
- the management function 910 checks whether or not the record of the statistical time management table 924 has been registered (step S261). If it is determined in step S261 that the record has been registered, the management function 910 proceeds to step S262. If it is determined that the record has not been registered, the management function 910 proceeds to step S263. When it is determined in step S261 that the record has been registered, the management function 910 transmits the queue statistical information to the distribution device 10 as information for the distribution device 10 to determine the distribution destination of the fixed-time event. (Step S262).
- the queue statistical information includes information of the analysis server column 9241 and the average estimated time column 9242 of the statistical time management table 924.
- the management function 910 sleeps (waits) for a certain time (step S263).
- the management function 910 may determine whether each analysis server 11 can satisfy the requested time based on the average estimated time, and may exclude the analysis server 11 determined not to satisfy the requested time from the queue statistical information. .
- the above is the queue statistical information transmission process.
- FIG. 25 shows the distribution order update process by the distribution apparatus 10.
- the distribution function 111 of the distribution apparatus 10 that has received the queue statistics information from the management server 9 by the above-described queue statistics information transmission process performs a distribution order update process for updating the distribution management table 121 based on the queue statistics information.
- the distribution function 111 receives queue statistical information from the management server 9 (step S271). Next, the distribution function 111 updates the distribution management table 121 based on the queue statistical information (step S272). Here, the distribution function 111 determines the priority of the analysis server 11 that is a distribution destination of the fixed-time event based on the average estimated time for each analysis server 11 indicated in the queue statistical information, and the distribution management table 121. In the distribution order column 1212 corresponding to the fixed time type. Further, the distribution function 111 registers the time when the record is updated in the update time column 1213. The above is the distribution order update process. In addition, the distribution function 111 of the distribution apparatus 10 that has received the distribution number information from the management server 9 by the above-described distribution number information transmission process registers the distribution number indicated in the distribution number information in the distribution number management table 721.
- FIG. 26 shows event transmission processing by the log output server 8.
- the node function 810 of the log output server 8 that has received the request from the client 7 performs a process corresponding to the request, and then performs an event transmission process of transmitting an access log indicating the process to the distribution apparatus 10 as an event.
- the node function 810 receives a request from the client 7 (step S291). Next, the node function 810 processes the request received in step S291 and generates its access log (step S292).
- the simple distribution function 811 selects the event transmission destination distribution device 10 in accordance with the simple distribution rule 820 from the distribution devices 10 registered in the distribution destination table 821 (step S293). Then, the node function 810 generates an event from the generated access log, and transmits the event to the event transmission destination (step S294).
- the event generated here is in accordance with the contents of the event type identification table 122 and has a protocol and a header corresponding to the event type. The above is the event transmission process.
- FIG. 27 shows distributed processing by the distributed device 10.
- the distribution function 111 of the distribution apparatus 10 determines the analysis server 11 that is the distribution destination and performs distributed processing for transmitting the event.
- the distribution function 111 receives an event from the log output server 8 (step S281). Next, the distribution function 111 identifies the event type of the received event based on the event type identification table 122. If it is a real-time type, the process proceeds to step S283, and if it is other than the real-time type, the process proceeds to step S287. (Step S282). Here, the distribution function 111 determines the protocol and header of the received event based on the information in the protocol column 1222 and the header column 1223 of the event type identification table 122, and the event type column 1221 corresponding to them. Get the event type. Based on the information, the determination is made in step S282.
- step S282 When the event type is not determined to be the real-time type in step S282 (determined to be the fixed time type) (step S282: N), the distribution function 111 refers to the distribution management table 121 and matches the event type column 1211. The distribution destination is acquired (step S287), and the process proceeds to step S285.
- the distribution function 111 selects, from among the plurality of analysis servers 11 shown in the distribution order of the distribution order column 1212, the analysis servers 11 corresponding to the number of distributions in the distribution number column 1214 as distribution destinations.
- step S282 When it is determined in step S282 that the event type is a real-time type (step S282: Y), the distribution function 111 acquires a distribution destination from the management server 9 (step S283).
- the distribution function 111 transmits a distribution destination query to the management server 9, receives a distribution destination response to the distribution destination query from the management server 9, and acquires the distribution destination included in the distribution destination response.
- the distribution function 111 transfers the event received in step S281 to the distribution destination of the analysis server 11 acquired in step S284 or step S287 (step S285). Then, the distribution function 111 transmits distribution destination information indicating the distribution destination to the management server 9 (step S286).
- the distribution destination information includes the event type and event type of the event transmitted by the distribution apparatus 10, identification information of the analysis server 11 at the distribution destination, and the like. The above is distributed processing.
- the distribution function 111 may select the analysis server 11 having a small number of events registered in the queue up to a predetermined selection number when distributing real-time events.
- the distribution apparatus 10 can select a distribution destination suitable for analyzing a fixed-time event. Further, the distribution apparatus 10 can acquire a distribution destination suitable for real-time event analysis from the management server 9.
- FIG. 28 shows distribution destination response processing by the management server 9.
- the management function 910 of the management server 9 that has received the distribution destination inquiry from the distribution apparatus 10 by the above-described distributed processing performs distribution destination response processing for transmitting a distribution destination response indicating the distribution destination to the distribution apparatus 10.
- the management function 910 transmits a distribution destination response indicating the distribution destination as a response to the distribution destination inquiry to the distribution apparatus 10 that is the transmission destination of the distribution destination inquiry (step S257).
- the management function 910 may determine whether each analysis server 11 can satisfy the requested time based on the estimated time, and may exclude the analysis server 11 determined not to satisfy the requested time from the distribution destination response. The above is the distribution destination response process.
- the management function 910 may repeat the process of step S252 for each distribution destination query each time, but considering the efficiency, the middle of the distribution destination response process is performed between a plurality of distribution destination queries. You can share it.
- FIG. 29 shows distribution destination registration processing by the management server 9.
- the management function 910 of the management server 9 that has received the distribution destination information from the distribution apparatus 10 by the above-described distribution processing performs a distribution destination registration process for registering the distribution destination information received from the distribution apparatus 10 in the time management table 923.
- the distribution destination information indicates a distribution destination by the distribution apparatus 10, and includes identification information of the analysis server 11 corresponding to the number of distributions from the top of the analysis server 11 indicated in the distribution order column 1212 of the distribution management table 121.
- the management function 910 receives the distribution destination information from the distribution apparatus 10 (step S231). Next, the management function 910 registers in the time management table 923 information such as the identification information and event type of the analysis server 11 to which the distribution apparatus 10 has distributed the event acquired in step S231 (step S232). The above is the distribution destination registration process.
- FIG. 30 shows an analysis process performed by the analysis server 11.
- the analysis function 1110 of the analysis server 11 that has received an event from the distribution apparatus 10 by the above-described distributed processing performs analysis processing for analyzing the event.
- the analysis function 1110 of the analysis server 11 receives an event from the distributed device 10 (step S301).
- the analysis function 1110 refers to the event type identification table 122 and identifies the received event type (step S302).
- the identification method here is the same method as step S282 of the distribution destination update process.
- the analysis function 1110 refers to the logic management table 1120 and determines the analysis logic corresponding to the identified event type (step S303). Next, the analysis function 1110 uses the queue management function 1111 to register the event received in step S301 in the queue management table 1121. After that, the analysis function 1110 analyzes in order from the higher-order event in the queue management table 1121, and generates analysis result information indicating the result (step S304).
- the analysis result information includes the event type and event type of the event to be analyzed, and the analysis value that is the result of the analysis. Then, the analysis function 1110 transmits the analysis result information to the determination device 12 (step S305).
- the analysis function 1110 analyzes from among the plurality of determination devices 12, for example, by carrying out accommodation distribution based on the hash value of the event content in the event content column 11213 of the queue management table 1121.
- the determination device 12 that is the transmission destination of the result information is selected.
- FIG. 31 shows the distribution number information update process by the determination device 12.
- the determination function 1210 of the determination apparatus 12 that has received the distribution number information from the management server 9 by the above-described distribution number information transmission process performs the distribution number information update process.
- the determination function 1210 receives the distribution number information from the management server 9 (step S311). Next, the determination function 1210 registers the distribution number for each event type in the distribution number management table 721 based on the received distribution number information (step S312). The above is the distributed number information update process.
- FIG. 32 shows an analysis result determination process performed by the determination device 12.
- the determination function 1210 of the determination device 12 that has received the analysis result information from the analysis server 11 by the above-described analysis processing performs analysis result determination processing for performing majority determination.
- the determination function 1210 receives the analysis result information from the analysis server 11 (step S321). Next, the determination function 1210 registers the received analysis result information in the analysis result management table 720 (step S322). Next, the determination function 1210 transmits analysis result reception information indicating reception of the analysis result information to the management server 9 (step S323).
- the analysis result reception information includes the event type, event type, and analysis value indicated in the analysis result information.
- the determination function 1210 acquires the number of distributions corresponding to the event type of the analysis result information from the distribution number management table 721, and analyzes the analysis result information having the same event type and the same event type from the analysis result management table 720. Among the indicated analysis values, the number of the same analysis values is counted.
- the determination function 1210 determines whether or not the number of specific analysis values has reached the majority of the target number of variances (more than half of the target number of variances) (step S324).
- the determination function 1210 moves the process to step S325 when the counted number of pieces of analysis result information reaches the majority of the number of distributions, and moves the process to step S321 when it does not reach the majority of the number of distributions.
- the determination function 1210 calculates the number of analysis result information for each analysis value from the analysis result information extracted in step S324. Count. Thereafter, the determination function 1210 makes a majority decision by comparing the number of pieces of analysis result information for each analysis value (step S325).
- the analysis value with the highest probability can be selected as the estimated value, and the reliability of the estimated value can be improved. Further, the determination function 1210 can determine the estimated value without waiting for all analysis result information, and can shorten the time until the estimated value is determined.
- the determination function 1210 counts the total number of analysis values, and determines that the analysis value with the larger total number is correct.
- the determination function 1210 may use an average value of all the analysis values as an estimated value or calculate a probability distribution of the analysis value. The expected value may be calculated from the probability distribution and used as the estimated value.
- the number of variances is an odd number.
- the number of distributions (the number of analysis servers 11) is an odd number.
- the determination function 1210 also has means for handling such a case.
- the determination function 1210 may select a predetermined designated value as an estimated value when a plurality of analysis values reach the same number. In this case, the management server 9 may transmit the specified value input from the administrator to the determination device 12. The determination function 1210 may give a predetermined weight to the total number of the plurality of values, and may select one of the plurality of values as the estimated value based on the weighted total number. In this case, the management server 9 may transmit the weight input from the administrator to the determination device 12.
- the determination function 1210 does not need to wait until the number of certain analysis values reaches a majority of the number of distributions in order to make a majority decision.
- the analysis value that has reached the determined number may be selected as an estimated value and transmitted to the AP server 13. Thereby, even if there are many dispersion
- the management server 9 may transmit the determination number input from the administrator to the determination device 12.
- the management function 910 of the management server 9 that has received the analysis result reception information from the determination device 12 by the analysis result determination process described above registers the analysis result reception information in the time management table 923 and the statistical time management table 924. Perform the registration process.
- the analysis result reception information indicates a reception status of the analysis result information by the determination device 12 and includes, for example, an analysis server column 12203 of the analysis result management table 720, an event type 12204, and the like.
- the management function 910 receives the analysis result reception information from the determination device 12 (step S241). Next, the management function 910 registers information based on the analysis result reception information acquired in step S241 in the time management table 923 and the statistical time management table 924. (Step S242).
- the management function 910 creates a record of the time management table 923, and shows the identification information, event type, and event type of the analysis server 11 indicated in the analysis result reception information, and the analysis server column 9231 and event type column of the record. 9233 and the event type column 9234, respectively. Furthermore, the management function 910 registers that the information is received from the determination device 12 in the reception source column 9232. Furthermore, the management function 910 registers the update time of the record in the update time column 9235 of the record.
- the management function 910 registers in the time management table 923 based on the update time of the record registered in the time management table 923 based on the distribution destination information (step S232 of the distribution destination registration processing) and the corresponding analysis result reception information.
- the difference from the update time of the recorded record (step S242 of the analysis result registration process) is calculated as the estimated time, and is registered in the estimated time column 9235 of the record.
- the management function 910 calculates a new total time by adding the calculated estimated time to the total time of the record corresponding to the analysis server 11 and the event type analysis logic of the record in the statistical time management table 924.
- the total time column 9244 of the record is updated.
- the management function 910 calculates a new counter value by adding 1 to the counter value of the record, and updates the counter column 9245 of the record.
- the management function 910 calculates the average estimated time by dividing the total time of the record by the counter value, and updates the average estimated time column 9242 of the record. The above is the analysis result registration process.
- the management server 9 does not perform communication with the analysis server 11, but analyzes the event after the event is input to the analysis server 11 from communication with the distribution device 10 and the determination device 12.
- the time required for completion can be estimated. This time is used for selecting a real-time type distribution destination as an estimated time. Also, the management server 9 calculates the average value of the estimated time and transmits it to the distribution device 10, so that it is not necessary to transmit the queue status for each event to the distribution device 10, and the load on the network can be suppressed. .
- This average value is used as an average estimated time for selecting a fixed time type distribution destination.
- analysis is performed on a log generated by a client access or the like, and the result is used by an application, and a method for determining a distribution destination is selected according to a request time by an AP.
- a method for determining a distribution destination is selected according to a request time by an AP.
- the AP server 13 there may be a mixture of APs that request to obtain an analysis result within a short request time and APs that only need to obtain an analysis result within a longer request time.
- the event type corresponding to the request time determined by the AP is determined, the analysis server for performing analysis is selected according to the event type, and the load of analysis is distributed. Analysis can be performed to meet each application requirement.
- the distribution according to the request time determined by the AP is not considered. Therefore, it cannot be efficiently distributed according to the type of event while satisfying the required time.
- the same analysis is distributed by a plurality of analysis servers, the analysis is executed in parallel, and the estimated value is calculated from the plurality of analysis values, so that even if the reliability of the analysis server 11 is low, the estimation is performed.
- the reliability of the value can be increased.
- the analysis server 11 can use the hardware of a general-purpose PC (Personal Computer) or the analysis logic of OSS (Open Source Software).
- the above embodiment does not require a high-performance and expensive analysis server, can reduce the software cost for analysis, and can reduce the cost of the analysis system.
- the present embodiment it is possible to exchange some hardware or software of the plurality of analysis servers 11 without stopping the analysis by the analysis system. Even if the reliability of some of the plurality of analysis servers 11 decreases due to this exchange, the reliability of the estimated value can be maintained by calculating the estimated value from the plurality of analyzed values. Thereby, the upgrade of the analysis server 11 etc. can be performed easily.
- the management server 9 determines the state of the queue of the analysis server 11 from the time when the distribution destination information is received from the distribution device 10 and the time when the analysis result reception information is received from the determination device 12. A method of estimating 11 loads has been described.
- the management server 9 receives information indicating the queue status from the analysis server 11 to obtain a method for grasping the queue status of the analysis server 11, that is, acquires information indicating the load from the analysis server 11. The method of grasping the load will be described.
- FIG. 34 shows the configuration of the management server 9 of the second embodiment.
- the memory 907 of the management server 9 of the present embodiment newly stores an analysis server state management table 925.
- the time management table 923 may be omitted.
- the management function 910 of the management server 9 sends a queue information query to the analysis server 11.
- the analysis server 11 that has received the inquiry transmits queue information indicating the status of its own queue to the management server 9.
- the management function 910 that has received the queue information registers information based on the queue information in the analysis server state management table 925.
- the queue information includes, for example, the queue length of the analysis server 11 (the number of processing waiting events accumulated in the queue).
- FIG. 35 shows an analysis server state management table 925 in the management server 9 of the second embodiment.
- the analysis server state management table 925 has a record for each analysis server 11. Each record has an analysis server column 9251, a queue length column 9252, an estimated time column 9253, and an update time column 9254.
- the analysis server column 9251 identification information of the analysis server 11 is registered.
- the queue length column 9252 the queue length of the analysis server 11 is registered.
- the estimated time column 9253 the estimated time (waiting time) until the analysis of the event newly accumulated in the queue of the analysis server 11 is completed is registered. In other words, this estimated time is the sum of the analysis times of events registered in the current queue.
- the update time column 9254 the time when the record was updated is registered.
- the management function 910 that has received the queue information registers in the analysis server state management table 925 the identification information of the analysis server 11 that is the transmission source, the queue length included in the queue information, and the time at which the queue information was received. Furthermore, the management function 910 calculates the estimated time by multiplying the analysis time in the analysis logic management table 921 by the queue length in the queue information received from the analysis server 11 and registers it in the analysis server state management table 925.
- the management server 9 acquires the distribution destination information and the analysis result reception information from the distribution device 10 and the determination device 12 arranged before and after the analysis server 11, and based on the acquired time
- the queue status of each analysis server 11, in short, the load is estimated, and the estimated result is stored as the estimated time in the time management table 923.
- the management server 9 makes an inquiry to the analysis server 11 to acquire the queue status, that is, load information, and stores the acquired result as the estimated time of the analysis server state management table 925.
- the same control as that of the first embodiment can be performed and the accuracy of the control is improved. be able to.
- the distribution apparatus 10 acquires the queue length of the analysis server 11 via the management server 9 and stores the acquired queue length in the distribution management table 121 to grasp the status of the queue of the analysis server 11. You may acquire load information. Furthermore, when the distribution apparatus 10 has a new event queue, for example, when distributing a fixed-time type event, the queue of the analysis server 11 has a certain amount based on the queue length information of the analysis server 11. The event is stored in the queue of the distribution device 10 so as not to exceed, and the distribution destination is determined. In short, the event transfer is controlled based on the acquired load information so that the load on each analysis server 11 does not exceed a predetermined amount. You may do it.
- a series of events in a certain log output server 8 is analyzed by the same analysis server 11.
- FIG. 36 shows the configuration of the dispersion apparatus 10 of the third embodiment.
- the memory 107 of the distribution apparatus 10 of the present embodiment further stores a session management table 124.
- FIG. 37 shows the session management table 124 in the distribution apparatus 10 of the third embodiment.
- the session management table 124 indicates a correspondence between a session established between the client 7 and the log output server 8 and the analysis server 11.
- the session management table 124 has a record for each session. Each record has a session ID column 1241, a distribution destination column 1242, an event type column 1243, and an event type column 1244.
- session ID column 1241 identification information of a session established between the client 7 and the log output server 8 is registered.
- distribution destination field 1242 identification information of the analysis server 11 that is the distribution destination to which the event in the session is transferred is registered.
- event type column 1243 the event type of the transferred event is registered.
- event type column 1244 the event type of the transferred event is registered.
- a session is one of access units and is a logical connection relationship for communication between the client 7 and the log output server 8.
- FIG. 38 shows a dispersion process performed by the dispersion apparatus 10 according to the third embodiment.
- the distributed processing of the present embodiment has steps S351 to S356 instead of step S283.
- the distribution function 111 of the distribution apparatus 10 determines whether or not the received event is the start of a new session (step S351).
- the event at the start of the session may include information indicating the start of the session.
- the distribution function 111 may determine whether or not the received event includes information indicating the start of a new session. Further, the distributed function 111 may determine whether or not there is a session record corresponding to the event type and event type of the received event in the session management table 124.
- the distributed function 111 determines whether the event type of the event is a real-time type in step S282 similar to the first embodiment. Determine whether or not. If it is determined that the event type is the real-time type, the distribution function 111 acquires the distribution destination from the management server 9 in step S283 similar to that in the first embodiment, and moves the process to step S353. If it is determined that the event type is not a real-time type (a constant time type), the distribution destination is determined based on the distribution management table 121 in step S287 similar to that in the first embodiment, and the process proceeds to step S353. Thereafter, the distributed function 111 creates a record of the started session in the session management table 124 (step S353), and moves the process to step S285 similar to the first embodiment.
- the distribution function 111 selects a record of the session corresponding to the received event in the session management table 124, and the record 2 is selected as a distribution destination (step S354).
- the distribution function 111 selects a record in which the event type and the event type indicated in the event type column 1243 and the event type column 1244 respectively match the received event in the session management table 124. Thereafter, the distribution function 111 selects the analysis server 11 of the distribution destination indicated in the distribution destination column 1242 of the selected record in the session management table 124.
- the distributed function 111 determines whether or not the received event is the end of the session (S355).
- the event at the end of the session may include information indicating the end of the session.
- the distribution function 111 may determine whether or not the received event includes information indicating the end of a new session.
- step S355: N When it is determined that the received event is not the end of the session (step S355: N), the distributed function 111 moves the process to step S285.
- step S355 When it is determined that the received event is the end of the session (step S355: Y), the distribution function 111 deletes the record of the selected session in the session management table 124, and the process is the same as in the first embodiment. To step S285. The subsequent processing is the same as in the first embodiment. The above is the distributed processing of this embodiment.
- the distribution function 111 may transmit an event in another series of operations such as a transaction to the same distribution destination instead of the session.
- the analysis server 11 can analyze a series of events. For example, in the log output server 8, when a certain event and a subsequent event are related, the distribution apparatus 10 transfers the event to the same analysis server 11, so that the analysis server 11 Analysis can be performed using those events. As a result, appropriate load distribution can be performed even when changes in values indicated by a plurality of events are analyzed, or when a series of operations performed by a client 7 is analyzed. For example, the present embodiment can be applied when analyzing a change in a certain stock price or when analyzing a series of operations in the client 7.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Computer Hardware Design (AREA)
- Software Systems (AREA)
- Debugging And Monitoring (AREA)
- Hardware Redundancy (AREA)
Abstract
Description
Claims (12)
- n個(nは3以上)の処理装置に接続された負荷分散装置と、
前記n個の処理装置に接続された処理結果判定装置と、
前記負荷分散装置及び前記処理結果判定装置に接続された管理装置と、を備え、
前記管理装置は、前記n個の処理装置の夫々について、過去の処理において要した時間情報を取得し、前記n個の処理装置の夫々が処理対象情報の処理に掛かる所要時間を、前記時間情報に基づいて推定して推定時間とし、前記推定時間及び処理装置の関係を示す推定時間情報を前記負荷分散装置へ送信し、
前記負荷分散装置は、前記推定時間情報に基づいて、前記処理対象情報を処理させる、m個(1<m<n)の処理装置を決定し、前記処理対象情報を、前記決定したm個の指定処理装置へ送信し、
前記m個の指定処理装置の夫々は、前記処理対象情報を処理することにより処理結果を算出し、前記処理結果を前記処理結果判定装置へ送信し、
前記処理結果判定装置は、前記m個の指定処理装置からm個の処理結果を夫々受信し、所定の方法により前記m個の処理結果から真の値を推定して推定値とし、前記推定値を処理する後段の処理装置へ前記推定値を送信する、処理制御システム。 - 前記処理対象情報の複数の分類の夫々に対して、通信装置による前記処理対象情報の送信から前記処理装置による前記推定値の受信までの時間の上限が予め定められており、
前記管理装置は、前記複数の分類の夫々に対応する推定時間情報を前記負荷分散装置へ送信し、
前記負荷分散装置は、前記処理対象情報の分類を判定し、前記判定された分類に対応する推定時間情報に基づいて前記m個の指定処理装置を決定する、請求項1に記載の処理制御システム。 - 前記処理結果判定装置は、処理結果毎に受信された数をカウントし、前記m個の処理結果の中で、最も多く受信された処理結果を前記推定値として選択する、請求項2に記載の処理制御システム。
- 前記管理装置は、前記指定処理装置の数を示す分散数mを前記処理結果判定装置へ送信し、
前記処理結果判定装置は、前記最も多く受信された処理結果の数が、前記分散数の半数を超えた場合、前記最も多く受信された処理結果を前記推定値として選択し、前記推定値を前記処理装置へ送信する、請求項3に記載の処理制御システム。 - 前記負荷分散装置は、前記処理対象情報を前記m個の指定処理装置へ送信したことを示す送信情報を前記管理装置へ送信し、
前記処理結果判定装置は、前記m個の指定処理装置の夫々から処理結果を受信したことを示す受信情報を前記管理装置へ送信し、
前記管理装置は、前記送信情報及び前記受信情報に基づいて前記時間情報を更新する、請求項2に記載の処理制御システム。 - 前記管理装置は、前記時間情報に基づいて過去の所要時間に関する所要時間情報を記憶し、前記所要時間情報に基づいて前記過去の所要時間の平均値を算出して前記推定時間とし、当該推定時間を、前記判定された分類において、前記指定処理装置の決定に用いる、請求項5に記載の処理制御システム。
- 前記管理装置は、前記n個の処理装置の夫々の前記推定時間を示す前記推定時間情報と、前記指定処理装置の数を示す分散数mとを前記負荷分散装置へ送信し、
前記負荷分散装置は、前記判定された分類において、前記n個の処理装置の中から前記推定時間が短い順に前記m個の指定処理装置を選択する、請求項6に記載の処理制御システム。 - 前記負荷分散装置は、前記推定時間情報の問合せを前記管理装置へ送信し、
前記管理装置は、前記問合せを受信した場合、前記n個の処理装置の中から前記推定時間が短い順に前記m個の指定処理装置を選択し、前記m個の指定処理装置を示す前記推定時間情報を前記負荷分散装置へ送信する、請求項7に記載の処理制御システム。 - 前記管理装置は、前記n個の処理装置に接続され、前記n個の処理装置の夫々のキューの状態を示すキュー情報を受信し、前記キュー情報に基づいて前記推定時間を推定する、請求項2に記載の処理制御システム。
- 前記通信装置は、互いに関連する複数の通信対象情報を前記負荷分散装置へ送信し、
前記負荷分散装置は、前記関連を示す関連情報を記憶し、前記関連情報に基づいて、前記複数の通信対象情報を同一の指定処理装置へ送信する、請求項2に記載の処理制御システム。 - n個(nは3以上)の処理装置を制御する処理制御方法であって、
前記n個の処理装置に接続された負荷分散装置と、前記n個の処理装置に接続された処理結果判定装置とに接続された管理装置が、前記n個の処理装置の夫々について、過去の処理において要した時間を示す時間情報を取得し、
前記管理装置が、前記n個の処理装置の夫々が処理対象情報の処理に掛かる所要時間を、前記時間情報に基づいて推定して推定時間とし、
前記管理装置が、前記推定時間及び処理装置の関係を示す推定時間情報を前記負荷分散装置へ送信し、
前記負荷分散装置が、前記推定時間情報に基づいて、前記処理対象情報を処理させる、m個(1<m<n)の処理装置を決定し、
前記負荷分散装置が、前記処理対象情報を前記決定したm個の指定処理装置へ送信し、
前記m個の指定処理装置の夫々が、前記処理対象情報を処理することにより処理結果を算出し、
前記m個の指定処理装置の夫々が、前記処理結果を前記処理結果判定装置へ送信し、
前記処理結果判定装置が、前記m個の指定処理装置からm個の指定処理結果を夫々受信し、
前記処理結果判定装置が、所定の方法により前記m個の処理結果から真の値を推定して推定値とし、
前記処理結果判定装置が、前記推定値を処理する後段の処理装置へ前記推定値を送信する、ことを備える処理制御方法。 - 負荷分散装置及び処理結果判定装置に接続された管理装置のコンピュータのための処理制御プログラムであって、
前記負荷分散装置及び前記処理結果判定装置に接続されたn個(nは3以上)の処理装置の夫々について、過去の処理において要した時間を示す時間情報を記憶し、
前記n個の処理装置の夫々が処理対象情報の処理に掛かる所要時間を、前記時間情報に基づいて推定して推定時間とし、
前記推定時間及び処理装置の関係を示す推定時間情報を前記負荷分散装置へ送信する
ことを前記コンピュータに実行させ、
前記負荷分散装置は、前記推定時間情報に基づいて、前記処理対象情報を処理させる、m個(1<m<n)の処理装置を決定し、前記処理対象情報を前記決定したm個の指定処理装置へ送信し、
前記m個の指定処理装置の夫々は、前記処理対象情報を処理することにより処理結果を算出し、前記処理結果を前記処理結果判定装置へ送信し、
前記処理結果判定装置は、前記m個の指定処理装置からm個の指定処理結果を夫々受信し、所定の方法により前記m個の処理結果から真の値を推定して推定値とし、前記推定値を処理する後段の処理装置へ前記推定値を送信する、処理制御プログラム。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14/772,723 US9501326B2 (en) | 2013-03-19 | 2014-03-04 | Processing control system, processing control method, and processing control program |
JP2015506685A JP6062034B2 (ja) | 2013-03-19 | 2014-03-04 | 処理制御システム、処理制御方法、および処理制御プログラム |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2013-056844 | 2013-03-19 | ||
JP2013056844 | 2013-03-19 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2014148247A1 true WO2014148247A1 (ja) | 2014-09-25 |
Family
ID=51579936
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2014/055475 WO2014148247A1 (ja) | 2013-03-19 | 2014-03-04 | 処理制御システム、処理制御方法、および処理制御プログラム |
Country Status (3)
Country | Link |
---|---|
US (1) | US9501326B2 (ja) |
JP (1) | JP6062034B2 (ja) |
WO (1) | WO2014148247A1 (ja) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018036830A (ja) * | 2016-08-31 | 2018-03-08 | 日本電信電話株式会社 | 処理システム、処理方法、処理サーバ、処理プログラム、評価サーバおよび評価プログラム |
JP2018045442A (ja) * | 2016-09-14 | 2018-03-22 | 株式会社東芝 | 中継装置、中継システム、中継プログラム、及び中継方法 |
JP2019075786A (ja) * | 2017-10-18 | 2019-05-16 | インフィネオン テクノロジーズ アーゲーInfineon Technologies Ag | データを処理するための方法および装置 |
CN113407322A (zh) * | 2021-06-21 | 2021-09-17 | 平安国际智慧城市科技股份有限公司 | 多终端的任务分配方法、装置、电子设备及可读存储介质 |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6610100B2 (ja) * | 2015-09-07 | 2019-11-27 | 富士通株式会社 | ログ分析方法、プログラム及び情報処理装置 |
JP6399127B2 (ja) * | 2017-03-08 | 2018-10-03 | 日本電気株式会社 | システム管理装置、システム管理方法、プログラム、情報処理システム |
US11734131B2 (en) * | 2020-04-09 | 2023-08-22 | Micron Technology, Inc. | Memory device having redundant media management capabilities |
CN111858043B (zh) * | 2020-07-10 | 2024-03-22 | 海尔优家智能科技(北京)有限公司 | 服务请求的处理方法及装置、存储介质、电子装置 |
EP4148581B1 (en) * | 2021-09-10 | 2023-08-30 | Axis AB | Verification of updated analytical procedures in monitoring systems |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2008152618A (ja) * | 2006-12-19 | 2008-07-03 | Fujitsu Ltd | ジョブ割当プログラム、方法及び装置 |
JP2009230581A (ja) * | 2008-03-24 | 2009-10-08 | Nippon Telegr & Teleph Corp <Ntt> | バッチジョブ制御システム、管理ノード、およびバッチジョブ制御方法 |
US20120084788A1 (en) * | 2010-10-05 | 2012-04-05 | Fujitsu Limited | Complex event distributing apparatus, complex event distributing method, and complex event distributing program |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07306794A (ja) | 1994-05-12 | 1995-11-21 | Mitsubishi Electric Corp | 分散システム及び分散システムの高信頼化方法 |
JP2003029989A (ja) * | 2001-07-16 | 2003-01-31 | Matsushita Electric Ind Co Ltd | 分散処理システムおよびジョブ分散処理方法 |
US7500133B2 (en) * | 2004-12-28 | 2009-03-03 | Sap Ag | Connection manager for handling message oriented protocol-based requests |
JP2006259812A (ja) | 2005-03-15 | 2006-09-28 | Hitachi Ltd | 動的キュー負荷分散方法、システム及びプログラム |
JP2007249491A (ja) * | 2006-03-15 | 2007-09-27 | Fujitsu Ltd | マルチサーバ環境においてバッチジョブを分散させるプログラム、装置、および方法 |
US8434085B2 (en) * | 2008-05-09 | 2013-04-30 | International Business Machines Corporation | Scalable scheduling of tasks in heterogeneous systems |
US8321870B2 (en) * | 2009-08-14 | 2012-11-27 | General Electric Company | Method and system for distributed computation having sub-task processing and sub-solution redistribution |
US9110724B2 (en) * | 2011-02-08 | 2015-08-18 | Microsoft Technology Licensing, Llc | Selecting computing nodes in cloud service using replication topologies |
-
2014
- 2014-03-04 US US14/772,723 patent/US9501326B2/en active Active
- 2014-03-04 WO PCT/JP2014/055475 patent/WO2014148247A1/ja active Application Filing
- 2014-03-04 JP JP2015506685A patent/JP6062034B2/ja not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2008152618A (ja) * | 2006-12-19 | 2008-07-03 | Fujitsu Ltd | ジョブ割当プログラム、方法及び装置 |
JP2009230581A (ja) * | 2008-03-24 | 2009-10-08 | Nippon Telegr & Teleph Corp <Ntt> | バッチジョブ制御システム、管理ノード、およびバッチジョブ制御方法 |
US20120084788A1 (en) * | 2010-10-05 | 2012-04-05 | Fujitsu Limited | Complex event distributing apparatus, complex event distributing method, and complex event distributing program |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018036830A (ja) * | 2016-08-31 | 2018-03-08 | 日本電信電話株式会社 | 処理システム、処理方法、処理サーバ、処理プログラム、評価サーバおよび評価プログラム |
JP2018045442A (ja) * | 2016-09-14 | 2018-03-22 | 株式会社東芝 | 中継装置、中継システム、中継プログラム、及び中継方法 |
JP2019075786A (ja) * | 2017-10-18 | 2019-05-16 | インフィネオン テクノロジーズ アーゲーInfineon Technologies Ag | データを処理するための方法および装置 |
JP7446708B2 (ja) | 2017-10-18 | 2024-03-11 | インフィネオン テクノロジーズ アーゲー | データを処理するための方法および装置 |
CN113407322A (zh) * | 2021-06-21 | 2021-09-17 | 平安国际智慧城市科技股份有限公司 | 多终端的任务分配方法、装置、电子设备及可读存储介质 |
Also Published As
Publication number | Publication date |
---|---|
US9501326B2 (en) | 2016-11-22 |
JP6062034B2 (ja) | 2017-01-18 |
JPWO2014148247A1 (ja) | 2017-02-16 |
US20160011909A1 (en) | 2016-01-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6062034B2 (ja) | 処理制御システム、処理制御方法、および処理制御プログラム | |
CN109218355B (zh) | 负载均衡引擎,客户端,分布式计算***以及负载均衡方法 | |
US11252220B2 (en) | Distributed code execution involving a serverless computing infrastructure | |
US9965441B2 (en) | Adaptive coalescing of remote direct memory access acknowledgements based on I/O characteristics | |
US20180084041A1 (en) | Method and system for dynamically rebalancing client sessions within a cluster of servers connected to a network | |
CN102281190B (zh) | 负载均衡装置组网方法以及服务器、客户端接入方法 | |
CN102918813B (zh) | 用于数据负载均衡的设备和方法 | |
CN100556024C (zh) | 用于网络场通信控制的方法和网络场 | |
US20060179059A1 (en) | Cluster monitoring system with content-based event routing | |
CN101167054A (zh) | 用于在多个数据中心之间选择性卸载工作负荷的方法和装置 | |
US20140115183A1 (en) | Information processing method, recording medium, and information processing apparatus | |
JP6272190B2 (ja) | 計算機システム、計算機、負荷分散方法及びそのプログラム | |
US20200028795A1 (en) | Distributed multi-tenant network real-time model for cloud based enterprise resource planning solutions | |
CN114338695B (zh) | 基于多副本的领导者选举方法、装置、设备及存储介质 | |
CN103825963B (zh) | 虚拟服务迁移方法 | |
US11695700B2 (en) | Information processing apparatus, computer-readable recording medium storing overload control program, and overload control method | |
US11558263B2 (en) | Network device association with network management system | |
CN111447282B (zh) | 用于确定传输路径的方法和装置 | |
US11233886B2 (en) | Storage medium and packet analyzing device | |
CN109445934A (zh) | 查询请求的分配方法及*** | |
US10148585B2 (en) | Communication control method, information processing apparatus, and storage medium | |
KR102642396B1 (ko) | 제한된 gpu리소스를 사용한 딥러닝 추론 모델을 위한 배치 스케줄링 장치 | |
US9128774B2 (en) | Information processing system for data transfer | |
CN112738193B (zh) | 云计算的负载均衡方法及装置 | |
US10834005B2 (en) | Buffer shortage management system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 14770196 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2015506685 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 14772723 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 14770196 Country of ref document: EP Kind code of ref document: A1 |