Automated scaling of an application and its support components

• US 8,706,852 B2
• Filed: 08/23/2011
• Issued: 04/22/2014
• Est. Priority Date: 08/23/2011
• Status: Active Grant

• Current Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Original Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Inventors: Kunze, Tobias, Ranganathan, Ram
• Primary Examiner(s): Avellino, Joseph E
• Assistant Examiner(s): KHAN, AFTAB N
• Application Number:
  US13/216,009
• Publication Number:
  US 20130054776A1
• Time in Patent Office: 973 Days
• Field of Search: 709223-229, 718/104, 719/328, 370/252, 370/468
• US Class Current: 709/221
• CPC Class Codes: G06F 2209/508 Monitor
• CPC Class Codes: G06F 9/5061 Partitioning or combining of resources

Abstract

An execution environment provides a plurality of components that support execution of an application, wherein the plurality of components and the application are hosted by a plurality of nodes arranged in one or more clusters. The execution environment monitors runtime behavior of the application and the plurality of components, and generates a scaling event based on a comparison of the runtime behavior to one or more scaling policies. The execution environment automatically scales at least one of the application or a component of the plurality of components if the scaling event does not violate a constraint, wherein the scaling comprises adding a new node to at least one cluster of the one or more clusters or removing a node from at least one cluster of the one or more clusters based on the scaling event.

Claims

1. A method comprising:

providing a plurality of components that support execution of an application, wherein the plurality of components and the application are hosted by a plurality of nodes arranged in a plurality of clusters, wherein the plurality of clusters have a plurality of scaling polices, wherein a first scaling policy in the plurality of scaling policies applies to a first cluster in the plurality of clusters, and wherein a second scaling policy in the plurality of scaling policies applies to a second cluster in the plurality of cluster;

monitoring, at the plurality of nodes, a first runtime behavior and a second runtime behavior of the application and the plurality of components, wherein the first runtime behavior occurs in the first cluster and the second runtime behavior occurs in the second cluster;

generating, by a first processing device in the first cluster, a first scaling event in view of a comparison of the first runtime behavior of the application or one of the plurality of components in the first cluster to the first scaling policy;

generating, by a second processing device in the second cluster, a second scaling event in view of a comparison of the second runtime behavior of the application or one of the plurality of components in the second cluster to the second scaling policy;

performing conflict resolution between the first scaling event and the second scaling event to determine which of the first scaling event or the second scaling event to process in view of determining that processing both the first scaling event and the second scaling event causes at least one of the first scaling event or the second scaling event to violate a constraint on resources within the first cluster or the second cluster; and

automatically scaling at least one of the application or a component in the plurality of components in view of processing the first scaling event or the second scaling event, wherein the scaling comprises adding a new node to the first cluster or the second cluster or removing an existing node from the first cluster or the second cluster in view of processing the first scaling event or the second scaling event.

2. The method of claim 1, wherein monitoring the first runtime behavior and the second runtime behavior comprises:

gathering metrics associated with resource utilization and performance of each of the application and the plurality of components; and

storing the metrics.

3. The method of claim 1, wherein each of the plurality of nodes is a virtual machine, and wherein adding the new node to the first cluster or the second cluster comprises:

instructing an underlying infrastructure to allocate a new virtual machine;

loading at least one of the application or the component onto the new virtual machine; and

adjusting a load balancing of the first cluster or the second cluster to utilize the new node.

4. The method of claim 1, further comprising:

managing a plurality of connections between the plurality of components and between the application and the plurality of components; and

updating at least one of the plurality of connections after performing the scaling.

5. The method of claim 4, wherein the plurality of connections are managed as publish-subscribe services, wherein a first endpoint of a connection in the plurality of connections is a publisher of the connection and a second endpoint of the connection is a subscriber of the connection, and wherein the method further comprises:

monitoring an address of the first endpoint; and

reporting changes in the address of the first endpoint to the second endpoint to maintain the connection.

6. The method of claim 1, further comprising:

generating a plurality of provisioning groups, wherein each of the plurality of provisioning groups comprises at least one of the application or one of the plurality of components, and wherein each of the plurality of provisioning groups further comprises a distinct cluster in the plurality of clusters to host the provisioning group.

7. The method of claim 6, wherein the first scaling event is associated with a first provisioning group in the plurality of provisioning groups and is generated in view of the comparison of the first runtime behavior for the first provisioning group to the first scaling policy for the first provisioning group, and wherein the new node is added to the first cluster for the first provisioning group.

8. The method of claim 6, wherein the second scaling event is associated with a second provisioning group in the plurality of provisioning groups and is generated in view of the comparison of the second runtime behavior for the second provisioning group to the second scaling policy for the second provisioning group, and wherein the existing node is removed from the second cluster for the second provisioning group.

9. The method of claim 6, further comprising:

modifying the plurality of provisioning groups during runtime of the application by removing the application or the component from a first provisioning group in the plurality of provisioning groups and adding the removed application or the removed component to a second provisioning group in the plurality of provisioning groups.

10. A non-transitory computer readable storage medium having instructions stored therein which, when executed by a processing device, cause the processing device to perform operations comprising:

providing a plurality of components that support execution of an application, wherein the plurality of components and the application are hosted by a plurality of nodes arranged in a plurality of clusters, wherein the plurality of clusters have a plurality of scaling polices, wherein a first scaling policy in the plurality of scaling policies applies to a first cluster in the plurality of clusters, and wherein a second scaling policy in the plurality of scaling policies applies to a second cluster in the plurality of cluster;

monitoring, at the plurality of nodes, a first runtime behavior and a second runtime behavior of the application and the plurality of components, wherein the first runtime behavior occurs in the first cluster and the second runtime behavior occurs in the second cluster;

generating, by a first processing device in the first cluster, a first scaling event in view of a comparison of the first runtime behavior of the application or one of the plurality of components in the first cluster to the first scaling policy;

generating, by a second processing device in the second cluster, a second scaling event in view of a comparison of the second runtime behavior of the application or one of the plurality of components in the second cluster to the second scaling policy;

performing conflict resolution between the first scaling event and the second scaling event to determine which of the first scaling event or the second scaling event to process in view of determining that processing both the first scaling event and the second scaling event causes at least one of the first scaling event or the second scaling event to violate a constraint on resources within the first cluster or the second cluster; and

automatically scaling at least one of the application or a component in the plurality of components in view of processing the first scaling event or the second scaling event, wherein the scaling comprises adding a new node to the first cluster or the second cluster or removing an existing node from the first cluster or the second cluster in view of processing the first scaling event or the second scaling event.

11. The non-transitory computer readable storage medium of claim 10, wherein monitoring the first runtime behavior and the second runtime behavior comprises:

gathering metrics associated with resource utilization and performance of each of the application and the plurality of components; and

storing the metrics.

12. The non-transitory computer readable storage medium of claim 10, wherein each of the plurality of nodes is a virtual machine, and wherein adding the new node to the first cluster or the second cluster comprises:

instructing an underlying infrastructure to allocate a new virtual machine;

loading at least one of the application or the component onto the new virtual machine; and

adjusting a load balancing of the first cluster or the second cluster to utilize the new node.

13. The non-transitory computer readable storage medium of claim 10, the operations further comprise:

managing a plurality of connections between the plurality of components and between the application and the plurality of components; and

updating at least one of the plurality of connections after performing the scaling.

14. The non-transitory computer readable storage medium of claim 13, wherein the plurality of connections are managed as publish-subscribe services, wherein a first endpoint of a connection in the plurality of connections is a publisher of the connection and a second endpoint of the connection is a subscriber of the connection, and wherein the operations further comprise:

monitoring an address of the first endpoint; and

reporting changes in the address of the first endpoint to the second endpoint to maintain the connection.

15. The non-transitory computer readable storage medium of claim 10, the operations further comprise:

generating a plurality of provisioning groups, wherein each of the plurality of provisioning groups comprises at least one of the application or one of the plurality of components, and wherein each of the plurality of provisioning groups further comprises a distinct cluster in the plurality of clusters to host the provisioning group.

16. The non-transitory computer readable storage medium of claim 15, wherein the first scaling event is associated with a first provisioning group in the plurality of provisioning groups and is generated in view of the comparison of the first runtime behavior for the first provisioning group to the first scaling policy for the first provisioning group, and wherein the new node is added to the first cluster for the first provisioning group.

18. The non-transitory computer readable storage medium of claim 15, the operations further comprise:

modifying the plurality of provisioning groups during runtime of the application by removing the application or the component from a first provisioning group in the plurality of provisioning groups and adding the removed application or the removed component to a second provisioning group in the plurality of provisioning groups.

17. The non-transitory computer readable storage medium of claim l5, wherein the second scaling event is associated with a second provisioning group in the plurality of provisioning groups and is generated in view of the comparison of the second runtime behavior for the second provisioning group to the second scaling policy for the second provisioning group, and wherein the existing node is removed from the second cluster for the second provisioning group.

19. A system comprising:

a plurality of storage devices to store and provide an application and a plurality of components that support execution of the application; and

a plurality of processing devices having a plurality of nodes that host the plurality of components and the application, wherein the plurality of nodes are arranged in a plurality of clusters, wherein the plurality of clusters have a plurality of scaling polices, wherein a first scaling policy in the plurality of scaling policies applies to a first cluster in the plurality of clusters, and wherein a second scaling policy in the plurality of scaling policies applies to a second cluster in the plurality of cluster, the plurality of processing devices to;

monitor a first runtime behavior and a second runtime behavior of the application and the plurality of components, wherein the first runtime behavior occurs in the first cluster and the second runtime behavior occurs in the second cluster;

generate, by a first processing device in the plurality of processing devices in the first cluster, a first scaling event in view of a comparison of the first runtime behavior of the application or one of the plurality of components in the first cluster to the first scaling policy;

generate, by a second processing device in the plurality of processing devices in the second cluster, a second scaling event in view of a comparison of the second runtime behavior of the application or one of the plurality of components in the second cluster to the second scaling policy;

perform conflict resolution between the first scaling event and the second scaling event to determine which of the first scaling event or the second scaling event to process in view of a determination that processing both the first scaling event and the second scaling event causes at least one of the first scaling event or the second scaling event to violate a constraint on resources within the first cluster or the second cluster; and

automatically scale at least one of the application or a component in the plurality of components by addition of a new node to the first cluster or the second cluster or removal of an existing node from the first cluster or the second cluster in view of the first scaling event or the second scaling event.

20. The system of claim 19, wherein the plurality of processing devices is to monitor the first runtime behavior and the second runtime behavior by collection of metrics associated with resource utilization and performance of each of the application and the plurality of components and storage of the metrics.