Efficient cache eviction and insertions for sustained steady state performance
A distributed metadata cache for a distributed object store includes a plurality of cache entries, an active-cache-entry set and an unreferenced-cache-entry set. Each cache entry includes information relating to whether at least one input/output (IO) thread is referencing the cache entry and information relating to whether the cache entry is no longer referenced by at least one IO thread. Each cache entry in the active-cache-entry set includes information that indicates that at least one IO thread is actively referencing the cache entry. Each cache entry in the unreferenced-cache-entry set is eligible for eviction from the distributed metadata cache by including information that indicates that the cache entry is no longer actively referenced by an IO thread.
1. A distributed metadata cache comprising:
a cache entry including a cache header, a first link to a first input/output (I/O) thread that references the cache entry, and a second link to a second I/O thread that is in a different referencing state than the first I/O thread,
wherein the cache header contains an identifier, and
wherein the different reference state indicates a decrease in hits of the cache entry.
2. The distributed metadata cache of claim 1 , wherein the first link to the first I/O thread is part of a set in a doubly linked list.
3. The distributed metadata cache of claim 2 , wherein the links in the set include previous and next pointers.
4. The distributed metadata cache of claim 1 , wherein the second link to the second I/O thread is part of a set in a doubly linked list.
5. The distributed metadata cache of claim 1 , wherein the first link to the first I/O thread is part of a set in a searchable-tree data structure.
6. The distributed metadata cache of claim 1 , wherein the distributed metadata cache maintains a reference counter accounting for a number of I/O threads that reference the cache entry.
7. The distributed metadata cache of claim 6 , wherein the I/O threads each release their respective cache entry.
8. The distributed metadata cache of claim 7 , wherein a determination that the cache entry is eligible for eviction is based on I/O characteristics of an application using the cache entry.
9. The distributed metadata cache of claim 8 , wherein a determination that the cache entry is eligible for eviction is based on a rate of consumption of free resources by the application.
10. A method of managing a distributed metadata cache comprising:
assigning to a cache entry a cache header, a first link to a first input/output (I/O) thread that references the cache entry, and a second link to a second I/O thread that is in a different referencing state than the first I/O thread,
wherein the cache header contains an identifier, and
wherein the different reference state indicates a decrease in hits of the cache entry.
11. The method of managing a distributed metadata cache of claim 10 , wherein the first link to the first I/O thread is part of a set in a doubly linked list.
12. The method of managing a distributed metadata cache of claim 11 , wherein the links in the set include previous and next pointers.
13. The method of managing a distributed metadata cache of claim 10 , wherein the second link to the second I/O thread is part of a set in a doubly linked list.
14. The method of managing a distributed metadata cache of claim 10 , wherein the first link to the first I/O thread is part of a set in a searchable-tree data structure.
15. The method of managing a distributed metadata cache of claim 10 further comprising:
maintaining a reference counter accounting for a number of I/O threads that reference the cache entry.
16. The method of managing a distributed metadata cache of claim 15 , wherein the I/O threads each release their respective cache entry.
17. The method of managing a distributed metadata cache of claim 16 further comprising:
determining that the cache entry is eligible for eviction based on I/O characteristics of an application using the cache entry.
18. The method of managing a distributed metadata cache of claim 17 , wherein the determining that the cache entry is eligible for eviction is based on a rate of consumption of free resources by the application.