Sharing local cache from a failover node
View Patent ↗Sharing local cache from a failover node, including: determining, by a managing compute node, whether a first compute node and a second compute node each have a local cache, where the second compute node is a mirrored copy of the first compute node; responsive to determining that the first compute node and the second compute node each have a local cache, combining, by the managing compute node, local cache on the first compute node and local cache on the second compute node into unified logical cache; receiving, by the managing compute node, a memory access request; and sending, by the managing compute node, the memory access request to an appropriate local cache in the unified logical cache.
1. A method comprising:
by program instructions on a computing device,
determining whether a first compute node and a failover compute node each have an internal cache, wherein the failover compute node is functioning as a mirrored copy of the first compute node;
responsive to determining that the first compute node and the failover compute node each have an internal cache, combining into a unified logical cache, a first internal cache of the first compute node and a second internal cache of the failover compute node;
receiving a memory access request; and
sending the memory access request to one of the internal caches in the unified logical cache including:
determining whether the unified logical cache includes data requested by the memory access request; and
responsive to determining that the unified logical cache includes data requested by the memory access request:
servicing the memory access request using non-cache memory on the first compute node, including reading data stored at a location in non-cache memory specified in the memory access request; and
writing the data to a location in the unified logical cache in dependence upon a predetermined cache replacement policy.
2. The method of claim 1 wherein writing the data to a location in the unified logical cache in dependence upon a predetermined cache replacement policy further comprises selecting one of the internal caches for receiving the data.
3. The method of claim 1 wherein the memory access request includes a memory address specifying a location in memory to read data from.