Apparatus, system and method for providing a persistent level-two cache
Aspects of the present disclosure involve a level-two persistent cache. In various aspects, a solid-state drive is employed as a level-two cache to expand the capacity of existing caches. In particular, any data that is scheduled to be evicted or otherwise removed from a level-one cache is stored in the level-two cache with corresponding metadata in a manner that is quickly retrievable.
1. A processor-based method for maintaining a cache, the method comprising:
evicting a data block from a level-one cache to a level-two cache, wherein the level-one cache comprises a volatile memory type, and wherein the level-two cache comprises a persistent memory type;
adding a first data block pointer into a ghost list in the level-one cache at a location identified by a ghost list index associated with the ghost list, wherein the first data block pointer references a location in the level-two cache where the data block is stored; and
adding, into a hash table:
a hash entry, at a location identified by a hash table index associated with the hash table, that includes:
an identifier of the data block; and
the ghost list index;
adding to a header in a data chunk in level-two cache;
a second data block pointer, wherein the second data block pointer references a location in the data chunk where the data block is stored;
the hash table index; and
the hash entry; and
detecting a loss of data in the level-one cache that includes an emptying of the level-one cache;
responding to the detecting a loss of data in the level-one cache by:
analyzing the header of the data chunk in the level-two cache to determine for the data block:
the hash entry;
the second data block pointer; and
the hash table index; and
repopulating the hash entry into the hash table at the location in the hash table identified by the hash table index;
generating the first data block pointer from the second data block pointer; and
repopulating the first data block pointer into the ghost list at the location in the ghost list identified by the ghost list index.
2. The processor-based method of claim 1 , wherein the data chunk further comprises a data section storing the data block and a header section storing the header and the header further comprises a magic number, and wherein analyzing the header includes analyzing the magic number to determine the data chunk is corrupted.
3. The processor-based method of claim 1 , wherein the level-one cache is an adaptive replacement cache.
4. The processor-based method of claim 1 , wherein the level-one cache and the level-two cache are part of a ZFS, wherein the level-one cache comprises dynamic random access memory and the level-two cache comprises one or more solid state drives.
5. The processor-based method of claim 4 , wherein the level-one cache is an adaptive replacement cache.
6. A processor-based system for maintaining a cache, the processor-based system comprising:
a level-two cache, wherein the level-two cache comprises a persistent memory type;
a level-one cache, wherein the level-one cache comprises a volatile memory type, comprising:
a ghost list comprising:
a ghost list index associated with the ghost list; and
a first data block pointer stored at a location identified by the ghost list index, wherein the first data block pointer references a location in the level-two cache where a data block is stored; and
a hash table comprising:
a hash table index associated with the hash table,
a hash entry, at the location identified by the hash table index, that includes:
an identifier of the data block, wherein the data block was evicted from the level-one cache to the level-two cache; and
the ghost list index;
a data chunk, located in the level-two cache, comprising:
the data block; and
a header comprising:
a second data block pointer, wherein the second data block pointer references a location of the data block in the data chunk;
the identifier of the data block;
the hash table index; and
the hash entry;
at least one processor to:
detect a loss of data in the level-one cache that includes an emptying of the level-one cache;
in response to the detecting the loss of data in the level-one cache:
analyze the header of the data chunk in the level-two cache to determine for the data block:
the second data block pointer;
the hash table index; and
the hash entry; and
repopulate the hash entry into the hash table at the location identified by the hash table index;
generate the first data block pointer from the second data block pointer; and
repopulate the first data block pointer into the ghost list at the location identified by the ghost list index.
7. The processor-based system of claim 6 , wherein the data chunk further comprises a data section storing the data block and a header section storing the header and the header further comprises a magic number, and wherein analyzing the header includes analyzing the magic number to determine the data chunk is corrupted.
8. The processor-based system of claim 6 , wherein the level-one cache is an adaptive replacement cache.
9. The processor-based system of claim 6 , wherein the level-on cache and the level-two cache are part of a ZFS, wherein the level-one cache comprises dynamic random access memory and the level-two cache comprises one or more solid state drives.
10. A non-transitory computer readable medium encoded with instructions for maintaining a cache executable by a processor, the instructions comprising:
evicting a data block from a level-one cache to a level-two cache, wherein the level-one cache comprises a volatile memory type, and wherein the level-two cache comprises a persistent memory type;
adding a first data block pointer into a ghost list in the level-one cache at a location identified by a ghost list index associated with the ghost list, wherein the first data block pointer references a location in the level-two cache where the data block is stored; and
adding, into a hash table:
a hash entry, at a location identified by a hash table index associated with the hash table, that includes:
an identifier of the data block; and
the ghost list index;
adding to a header in a data chunk in level-two cache;
a second data block pointer, wherein the second data block pointer references a location in the data chunk where the data block is stored;
the hash table index; and
the hash entry; and
detecting a loss of data in the level-one cache that includes an emptying of the level-one cache;
responding to the detecting a loss of data in the level-one cache by:
analyzing the header of the data chunk in the level-two cache to determine for the data block:
the hash entry;
the second data block pointer; and
the hash table index; and
repopulating the hash entry into the hash table at the location in the hash entry identified by the hash table index;
generating the first data block pointer from the second data block pointer; and
repopulating the first data block pointer into the ghost list at the location in the ghost list identified by the ghost list index.
11. The non-transitory computer readable medium of claim 10 , wherein the data chunk further comprises a data section storing the data block and a header section storing the header and the header further comprises a magic number, and wherein analyzing the header includes analyzing the magic number to determine the data chunk is corrupted.
12. The non-transitory computer readable medium of claim 10 , wherein the level-one cache is an adaptive replacement cache.
13. The non-transitory computer readable medium of claim 10 , wherein the level-on cache and the level-two cache are part of a ZFS, wherein the level-one cache comprises dynamic random access memory and the level-two cache comprises one or more solid state drives.
14. The non-transitory computer readable medium of claim 13 , wherein the level-one cache is an adaptive replacement cache.
15. The processor-based system of claim 9 , wherein the level-one cache is an adaptive replacement cache.