ZERO DOWNTIME MECHANISM FOR SOFTWARE UPGRADE OF A DISTRIBUTED COMPUTER SYSTEM

• US 20100162226A1
• Filed: 12/18/2008
• Published: 06/24/2010
• Est. Priority Date: 12/18/2008
• Status: Active Grant

• Current Assignee: SAP SE
• Original Assignee: SAP SE
• Inventors: Mitev, Dencho, Driesen, Volker, Borissov, Lazar, Mihalev, Ivan, Georgiev, Anton, Brodkorb, Thomas
• Granted Patent: US 9,229,707 B2
• Time in Patent Office: Days
• US Class Current: 717/173
• CPC Class Codes: G06F 8/65 Updates security arrangements therefor G06F21/57
• CPC Class Codes: G06F 8/656 while running
• CPC Class Codes: G06F 8/658 Incremental updates; Differential updates

Abstract

A system and a method for installing software upgrades in a distributed computer system with no downtime. In a distributed computer system with cluster architecture, a number of service requests are redirected from a first instance of the cluster to a second instance of the cluster for execution. An execution in progress of a service request is also redirected from the first instance of the cluster to the second instance of the cluster. The execution in progress proceeds from a state that is saved on a public store by the first instance of the cluster. A software patch is applied to one or more software components running on an application server node of the first instance of the cluster. A number of service requests are redirected from the second instance of the cluster to the first instance of the cluster for execution. An execution in progress of a service request is also redirected from the second instance of the cluster to the upgraded first instance of the cluster. The execution in progress proceeds from a state that is saved on the public store by the second instance of the cluster. The software patch is applies on the one or more software components running on an application server node of the second instance of the cluster. A number of new service requests are directed to the upgraded second instance of the cluster for processing.

Claims

1. A computer system comprising:

a first instance of a cluster to process a first plurality of service requests;

a second instance of the cluster to process a second plurality of service requests;

a dispatcher to route the first plurality of service requests to the first instance and the second plurality of service requests to the second instance of the cluster;

a patch tool to initiate an upgrade of a plurality of software components running on the cluster instances, the upgrade is performed in rolling mode with zero downtime for the cluster; and

a public store to store an execution code and related data for the plurality of software components running on the cluster instances, the related data comprising one of a state of execution and application data.

2. The system of claim 1, wherein an instance of the cluster comprises:

one or more application server nodes, each application server node to run the plurality of software components to process a plurality of service requests;

a communicator to receive the plurality of service requests from the dispatcher and to forward a service request from the plurality of service requests to one of the one or more application server nodes for processing, and

a control framework to start, stop and monitor the work of the one or more application servers and the communicator.

3. The system of claim 2, wherein each application server node comprises:

a services manager to provide running environment for a plurality of system software components and a plurality of application software components; and

a session manager to handle a plurality of user sessions and to store them on the public store, each of the plurality of user sessions store a state of an execution of one or more service requests initiated by a corresponding user;

a deploy controller to start, pause and stop an upgrade process for the application server node; and

a deploy service to apply one or more software upgrades to one or more application software components in the public store and to synchronize the running one or more application software components on the application server node with the upgrade.

4. The system of claim 3, wherein the service manager provides further to:

apply one or more software upgrades to one or more system software components in the public store and to synchronize the running one or more system software components on the application server node with the upgrade.

5. The system of claim 3, wherein the session manager provides further to:

read from the public store a plurality of user sessions stored from another service instance in order to provide failover for a plurality of service requests.

6. The system of claim 3, wherein the deploy controller provides further to:

receive from the patch tool a software patch comprising a plurality of software upgrades for the plurality of software components,

provide one or more software upgrades from the plurality of software upgrades to service manager for upgrading one or more system software components, and

provide one or more software upgrades from the plurality of software upgrades to the deploy service for upgrading one or more application software components.

7. The system of claim 2, wherein the communicator provides further to:

redirect a subset of the plurality of service requests to another instance of the cluster; and

forward one or more service requests from the subset of the service requests to one of the one or more application server nodes in order to test whether software upgrades are applied correctly.

8. A computerized method comprising:

preparing a software patch update for upgrading one or more software components from a plurality of software components running on a plurality of application server nodes of a cluster;

redirecting a plurality of service requests from a first instance of the cluster to a second instance of the cluster for execution, the plurality of service requests are executed by the plurality of software components;

redirecting an execution in progress of a service request from the first instance of the cluster to the second instance of the cluster, the execution in progress proceeds from a state that is saved on a public store by the first instance of the cluster; and

applying the software patch to the one or more software components running on an application server node of the first instance of the cluster.

9. The computerized method of claim 8 further comprising:

redirecting a plurality of service requests from the second instance of the cluster to the first instance of the cluster for execution, the plurality of service requests are executed by the plurality of software components including the upgraded one or more software components;

redirecting an execution in progress of a service request from the second instance of the cluster to the first instance of the cluster, the execution in progress proceeds from a state that is saved on the public store by the second instance of the cluster; and

applying the software patch to the one or more software components running on an application server node of the second instance of the cluster.

10. The method of claim 8, wherein preparing the software patch update comprises:

receiving at a patch tool a software patch comprising one or more software upgrades to be installed on the cluster;

connecting the patch tool to the first instance of the cluster; and

creating a backup copy of the first instance of the cluster comprising the plurality of software components running on the application server node of the first instance of the cluster.

11. The method of claim 8, wherein preparing the software patch update comprises:

identifying the one or more software components from the plurality of software components running on the plurality of application server nodes of the cluster that will be upgraded with the patch; and

ensuring a copy of an original installation of the identified one or more software component, the original installation comprises a current version of the one or more software components.

12. The method of claim 8, wherein redirecting the execution in progress of the service request comprises:

storing a user session to the public store by a session manager of the first instance of the cluster, the user session keeps a current state of the execution in progress of the service request;

reading the user session from the public store by a session manager of the second instance of the cluster; and

proceeding with the execution of the service request on the second cluster instance from the current state of the execution stored in the user session.

13. The method of claim 8, wherein applying the software patch comprises:

applying one or more software upgrades to an binary data of the one or more software components, the binary data is stored in the public store; and

synchronizing the one or more software components running on the application server node of the first instance of the cluster with the upgraded binary data.

14. The method of claim 8, wherein applying the software patch comprises:

routing a test service request from the redirected plurality of service requests to the first instance;

processing the test service request at the upgraded one or more software components at the first instance of the cluster; and

approving the patch or rolling back the patch, according to the result of the processing.

15. The method of claim 9, wherein applying the software patch comprises:

synchronizing the one or more software components running on the application server node of the second instance of the cluster with the upgraded binary data.

16. An article of manufacture having computer readable instructions stored thereon for execution by a processor to perform a method, the method comprising:

preparing a software patch update for upgrading one or more software components from a plurality of software components running on a plurality of application server nodes of a cluster;

redirecting a plurality of service requests from a first instance of the cluster to a second instance of the cluster for execution, the plurality of service requests are executed by the plurality of software components;

redirecting an execution in progress of a service request from the first instance of the cluster to the second instance of the cluster, the execution in progress proceeds from a state that is saved on a public store by the first instance of the cluster; and

applying the software patch to the one or more software components running on an application server node of the first instance of the cluster.

17. The article of manufacture of claim 16 having computer readable instructions stored thereon for execution by a processor to perform a method, the method further comprising:

redirecting a plurality of service requests from the second instance of the cluster to the first instance of the cluster for execution, the plurality of service requests are executed by the plurality of software components including the upgraded one or more software components;

redirecting an execution in progress of a service request from the second instance of the cluster to the first instance of the cluster, the execution in progress proceeds from a state that is saved on the public store by the second instance of the cluster; and

applying the software patch to the one or more software components running on an application server node of the second instance of the cluster.

18. The article of manufacture of claim 16, wherein redirecting the execution in progress of the service request comprises:

storing a user session to the public store by a session manager of the first instance of the cluster, the user session keeps a current state of the execution in progress of the service request;

reading the user session from the public store by a session manager of the second instance of the cluster; and

proceeding with the execution of the service request on the second cluster instance from the current state of the execution stored in the user session.

19. The article of manufacture of claim 16, wherein applying the software patch comprises:

applying one or more software upgrades to an binary data of the one or more software components, the binary data is stored in the public store; and

synchronizing the one or more software components running on the application server node of the first instance of the cluster with the upgraded binary data.

20. The article of manufacture of claim 19 having computer readable instructions stored thereon for execution by a processor to perform a method, the method further comprising:

routing a test service request from the redirected plurality of service requests to the first instance;

processing the test service request at the upgraded one or more software components at the first instance of the cluster; and

approving the patch or rolling back the patch, according to the result of the processing.