Distributed caching cluster management

• US 9,847,907 B2
• Filed: 11/26/2012
• Issued: 12/19/2017
• Est. Priority Date: 11/26/2012
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Shankaran, Nishanth, Shou, Dong, Arya, Rajat, Sbisa, Clint Joseph, Krishnamoorthy, Shyam, Thiruchi Loganathan, Rajaprabhu
• Primary Examiner(s): Joo, Joshua
• Application Number:
  US13/685,615
• Publication Number:
  US 20140149537A1
• Time in Patent Office: 1,849 Days
• Field of Search: None
• CPC Class Codes: H04L 41/0668 by dynamic selection of recovery network elements, e.g. replacement by the most appropriate element after failure
• CPC Class Codes: H04L 41/0816 the condition being an adaptation, e.g. in response to network events
• CPC Class Codes: H04L 67/568 Storing data temporarily at an intermediate stage, e.g. caching

Abstract

A management system may enable and monitor a cache or other cluster to make the cluster configuration-aware such that initialization and changes to the underlying structure of the cluster can be dynamically updated. For example, a distributed memory caching system may provide initial configuration to a client from a memory caching node referenced by an alias provided by a configuration endpoint. Updates of configuration may be retrieved from memory caching nodes, each storing current configuration of the cache cluster. A management system monitors changes to the cache cluster, such as provisioning of new caching nodes, and updates the configuration stored in the caching nodes for retrieval by a client.

Claims

1. A computer-implemented method for managing a distributed memory caching system, comprising:

managing, by a management system of the distributed memory caching system, a configuration endpoint of a cache cluster, the configuration endpoint associated with a configuration stored in a partitioned memory space of a set of memory caching nodes, wherein the management system is connected to the configuration endpoint and to the set of memory caching nodes by a network;

monitoring, utilizing the set of memory caching nodes in the cache cluster, for changes to the configuration made by other memory caching nodes in the set of memory caching nodes, each memory caching node of the set of memory caching nodes providing storage for a portion of data contained in the cache cluster and each of the set of memory caching nodes of the cache cluster comprising the configuration;

detecting that a new memory caching node has been added to the cache cluster, the detection that the new memory caching node has been added to the cache cluster being based at least in part on the monitoring by the set of memory caching nodes for configuration changes made by the other memory caching nodes;

provisioning, by the management system, the new memory caching node into the cache cluster;

updating, by the management system, the configuration within the partitioned memory space of at least one of the set of memory caching nodes based at least in part on provisioning the new memory caching node; and

updating, by the management system, the configuration endpoint with an alias to a memory caching node of the set of memory caching nodes based at least in part on updating the configuration stored in the partitioned memory space of the at least one of the set of memory caching nodes.

2. The computer-implemented method of claim 1:

wherein the monitoring for changes to the configuration made by the other memory caching nodes in the set of memory caching nodes further comprises detecting a failure of a failed memory caching node from the cache cluster;

wherein provisioning the new memory caching node further comprises causing a replacement memory caching node to be provisioned in place of the failed memory caching node; and

wherein updating the configuration within the partitioned memory space of the at least one of the set of memory caching nodes is based at least in part on the replacement of the failed memory caching node with the replacement memory caching node.

3. The computer-implemented method of claim 1:

wherein the monitoring for changes to the configuration made by the other memory caching nodes in the set of memory caching nodes further comprises detecting an indicator indicating a need of additional caching performance from the cache cluster;

wherein provisioning the new memory caching node into the cache cluster further comprises provisioning the new memory caching node in place of a replaced memory caching node that has less resources than the new memory caching node; and

wherein updating the configuration within the partitioned memory space of the at least one of the set of memory caching nodes is based at least in part on the replacement of the replaced memory caching node with the new memory caching node.

4. The computer-implemented method of claim 1, wherein updating the configuration endpoint with the alias to the memory caching node of the set of memory caching nodes is based at least in part on provisioning of the new memory caching node.

5. The computer-implemented method of claim 1, wherein provisioning the new memory caching node into the cache cluster further comprises constructing a workflow that includes the operations of:

provisioning the new memory caching node;

adding the new memory caching node to the set of memory caching nodes;

determining a new configuration; and

updating the configuration within the partitioned memory space of each of the set of memory caching nodes.

6. The computer-implemented method of claim 1, wherein the partitioned memory space of at least one of the set of memory caching nodes is excluded from being subject to eviction and the expiration rules defined by a caching protocol.

7. The computer-implemented method of claim 1, wherein the partitioned memory space of at least one of the set of memory caching nodes is configured to be accessed utilizing an extension of a caching protocol.

8. The computer-implemented method of claim 7, wherein the caching protocol comprises a memcached protocol.

9. The computer-implemented method of claim 7, wherein the caching protocol comprises at least one of an eviction or an expiration rule.

10. A computer-implemented method for managing a caching service, comprising:

managing, by a management system, a configuration endpoint of a cluster, the configuration endpoint associated with a configuration of the cluster stored in a partitioned memory space of a set of storage nodes, wherein the management system is connected to the configuration endpoint and to the set of storage nodes by a network;

monitoring, utilizing the set of storage nodes in the cluster, for configuration changes of the cluster made by other storage nodes of the set of storage nodes, the cluster comprising the set of storage nodes, each storage node of the set of storage nodes providing storage for a portion of data contained in a distributed data store;

enabling, by the management system, storage in each of the set of storage nodes, of the configuration for accessing data within the cluster;

detecting that the configuration has been updated, the detection being based at least in part on the monitoring by the set of storage nodes for the configuration changes of the cluster made by the other storage nodes;

as a result of the detection that the configuration be updated, updating the configuration within the partitioned memory space of at least one of the set of storage nodes;

updating, by the management system, the configuration stored in the partitioned memory space of one of the set of storage nodes based on monitoring changes to the cluster; and

updating, by the management system, the configuration endpoint with an alias to a storage node of the set of storage nodes based at least in part on updating, by the management system, the configuration stored in the partitioned memory space of one of the set of storage nodes.

11. The computer-implemented method of claim 10 further comprising modifying the cluster by removing a particular storage node from the cluster through the operations of:

de-provisioning the particular storage node; and

reallocating at least a portion of data contained in the particular storage node to at least one of a plurality of remaining storage nodes in the cluster.

12. The computer-implemented method of claim 10 further comprising modifying the cluster by adding a new storage node to the cluster through the operations of:

provisioning the new storage node; and

reallocating a sub-portion of the data contained in the distributed data store to the new storage node.

13. The computer-implemented method of claim 10 further comprising modifying the cluster by altering a characteristic of a particular storage node of the cluster.

14. The computer-implemented method of claim 13, wherein altering the characteristic of the particular storage node further comprises modifying a storage size of the particular storage node.

15. A computer system for managing a caching service, comprising:

one or more computing resources having one or more processors and memory including executable instructions that, when executed by the one or more processors, cause the one or more processors to implement at least;

a set of memory caching nodes forming a cache cluster that;

provide storage for a distributed data cache allocated across the cache cluster;

provide storage of a configuration for accessing data within the cache cluster, the configuration stored in a partitioned memory space of at least one of the set of memory caching nodes;

monitor for configuration changes made by other memory caching nodes in the cluster;

detect changes to the configuration based at least in part on the monitoring; and

update the configuration within the partitioned memory space of the at least one of the set of memory caching nodes based at least in part on detecting the changes to the configuration; and

a management system that;

manages a configuration endpoint of the cache cluster, the configuration endpoint associated with the configuration stored in the partitioned memory space of the at least one of the set of memory caching nodes;

monitors and provisions memory caching nodes;

updates the configuration stored in the partitioned memory space of one of the set of memory caching nodes, the updated configuration being based at least in part on alterations to the cache cluster caused by the management system; and

updates the configuration endpoint with an alias to a memory caching node of the set of memory caching nodes based at least in part on updating the configuration stored in the partitioned memory space of one of the set of memory caching nodes.

16. The computer system of claim 15, further comprising a data store that provides data for the distributed data cache that is allocated across the cache cluster.

17. The computer system of claim 16, wherein the data store is a relational database and wherein the distributed data cache includes responses to queries from the relational database.

18. The computer system of claim 15, the configuration endpoint configured to provide the alias to the memory caching node of the set of memory caching nodes from the cache cluster and the configuration endpoint configured to receive alias updates from the management system.

19. The computer system of claim 18, wherein the configuration endpoint has a static address that is resolvable to a particular memory caching node of the cache cluster.

20. The computer system of claim 15, further comprising a program execution environment in which the cache cluster and management system are allocated computing resources on which to execute.

21. One or more non-transitory computer-readable storage media having collectively stored thereon executable instructions that, when executed by one or more processors of a computer system, cause the computer system to at least:

provide a configuration endpoint of a cache cluster, the configuration endpoint associated with a configuration stored in a partitioned memory space of a set of memory caching nodes;

monitor, utilizing the set of memory caching nodes of the cache cluster, the cache cluster for indications of failure of one or more of the set of memory caching nodes, the set of memory caching nodes providing storage for a portion of data contained in a distributed data cache;

determine that an indicator of a particular memory caching node from the cache cluster has passed a failure threshold, the determination that the indicator has passed the failure threshold being based at least in part on the monitoring by the set of memory caching nodes;

cause the particular memory caching node to be removed from the cache cluster; and

update the configuration stored in the partitioned memory space of each of the set of memory caching nodes of the cache cluster based at least in part on the determination that the indicator of the particular memory caching node has passed the failure threshold, the update of the configuration stored in the partitioned memory space of each of the set of memory caching nodes causing the computer system to update the configuration endpoint with an alias to a memory caching node of the set of memory caching nodes.

22. The non-transitory computer-readable storage media of claim 21, wherein the executable instructions further comprise instructions that, when executed, cause the computer system to at least:

cause a new memory caching node to be provisioned and added to the cache cluster; and

update the configuration within at least one memory caching node from the cache cluster based at least in part on the provisioned new memory caching node.

23. The non-transitory computer-readable storage media of claim 22, wherein updating the configuration stored in the partitioned memory space of each of the set of memory caching nodes of the cache cluster further comprises updating the configuration such that the new memory caching node replaces the particular memory caching node removed from the cache cluster.

24. The non-transitory computer-readable storage media of claim 21, wherein the configuration stored in the partitioned memory space of each of the set of memory caching nodes of the cache cluster comprises connection information to the cache cluster.

25. The non-transitory computer-readable storage media of claim 21, wherein updating the configuration stored in the partitioned memory space of each of the set of memory caching nodes of the cache cluster causes the set of memory caching nodes to propagate the configuration using peer-to-peer updating.