Distributed processing system

Patent metadata

Field	Value
Publication number	US-2017123941-A1
Application number	US-201515311974-A
Country	US
Kind code	A1
Filing date	Apr 24, 2015
Priority date	May 19, 2014
Publication date	May 4, 2017
Grant date	—

Abstract

Provided is a distributed processing system in which migration evaluation is performed based on performance information, statistical information of each process of an application, and the number of non-completed processes of each process of an application for an application server in which an old application is being operated, a migration target server group is decided so that migration is completed in a shortest period of time, and migration to a new application is performed in a stepwise manner.

First claim

1 . A distributed processing system, comprising:
a plurality of application servers; and a management device, wherein the application server includes an application portion and a distributed execution platform portion, the application portion includes an adapter unit that receives an execution request of an application from a client terminal, and transmits a message of a processing request of the application to the distributed execution platform portion, and a processor unit that performs a process of the application in response to a request from the distributed execution platform, the distributed execution platform portion includes a dispatcher unit that holds the message transmitted from the adapter unit, and selects an application server that performs the process of the application requested through the message according to a routing strategy, and a statistical information storage unit that holds statistical information of each process of the application by the application server, and the management device includes a migration management unit that manages a migration status of the application for every two or more application servers, and a migration evaluating unit that decides a migration target server group based on performance information of the application server, statistical information of each process of the application, and the number of non-completed processes of each process calculated based on the number of messages held in the dispatcher unit for the application server in which the migration status is an old application operation state.

2 . The distributed processing system according to claim 1 ,
wherein after the process of the application is performed, when there is a process subsequent to the process, the processor unit transmits an execution request for the subsequent process to the adapter unit, when the execution request for the subsequent process is received, the adapter unit transmits a message of a processing request of the subsequent process to the distributed execution platform portion, the management device includes a definition storage unit that manages an execution order of the process and the subsequent process when the execution request of the application is received from the client terminal, and the migration evaluating unit calculates the number of non-completed processes by adding the number of messages held in the dispatcher unit to the number of processes subsequent to the process requested through the message.

3 . The distributed processing system according to claim 2 ,
wherein the migration evaluating unit generates a combination pattern of application servers in which the number of servers necessary for performing processes of the application corresponding to the number of non-completed processes is minimum based on the performance information, the statistical information, and the number of non-completed processes, and decides an application server included in a combination pattern in which an expected completion period of time of a process included in an application server not included in the combination pattern is shortest as the migration target server group.

4 . The distributed processing system according to claim 3 ,
wherein the adapter unit includes a first adapter that receives the execution request of the application from the client terminal and a second adapter that receives an execution request of a process from the processor unit, the management device includes an instructing unit that gives a prior blockage instruction to an application server included in the migration target server group decided by the migration evaluating unit, the dispatcher unit of the application server that has received the prior blockage instruction changes the routing strategy to select an application server other than the application server of the dispatcher unit when there are a plurality of candidates of an application server that performs the process, and the adapter unit of the application server that has received the prior blockage instruction stops the first adapter.

5 . The distributed processing system according to claim 4 , further comprising,
a server sharing device that manages states of a plurality of application servers, and notifies the application server of an application server that is able to perform the process of the application, the dispatcher unit of the application server that has received the prior blockage instruction gives a notification indicating that the application server of the dispatcher unit is unable to perform the process of the application to the server sharing device, the server sharing device updates the state of the application server, and transmits an updated executable application server list to another application server, and the dispatcher unit of the other application server selects an application server that performs the process according to the routing strategy among the received executable application server list.

6 . The distributed processing system according to claim 5 ,
wherein the management device includes a monitoring unit that monitors the number of messages held in the dispatcher unit of the application server that has received the prior blockage instruction from the instructing unit, and a migration unit that gives an instruction to perform application migration to an application server in which the number of messages monitored by the monitoring unit is zero, and the migration management unit updates the application server that has received the instruction to perform the application migration from the migration unit to a new application operation state.

7 . The distributed processing system according to claim 6 ,
wherein after the migration unit gives the instruction to perform the application migration to all application servers to which the instructing unit gives the prior blockage instruction, and the migration evaluating unit acquires the performance information of the application server, the statistical information of each process of the application, and the number of non-completed processes of each process calculated based on the number of messages held in the dispatcher unit for the application server in which the migration status is an old application operation state again, and decides a new migration target server group.

8 . The distributed processing system according to claim 7 ,
wherein the migration unit gives the instruction to perform the application migration to an application server in which the number of messages monitored by the monitoring unit is zero consecutively a predetermined number of times.

Assignees

• Hitachi Ltd

Inventors

• Yubamoto Jun
• Miyata Tatsuhiko

Classifications

• G06F11/00

  Error detection; Error correction; Monitoring (error detection, correction or monitoring in information storage based on relative movement between record carrier and transducer G11B20/18; monitoring, i.e. supervising the progress of recording or reproducing G11B27/36; in static stores G11C29/00) · CPC title

• G06F11/20

  using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements · CPC title

• G06F11/203 Primary

  using migration · CPC title

• G06F2201/805

  Real-time · CPC title

• G06F8/656

  while running · CPC title