System and method for managing large filesystem-based caches
View Patent ↗Embodiments disclosed herein utilize statistical approximations to manage large filesystem-based caches based on imperfect information. When removing entries from a large cache, which may have a million or more entries, the cache manager does not need to find the absolutely oldest entry that has been accessed the least recently. Instead, it suffices to find an entry that is older than most. In embodiments disclosed herein, statistical sampling of the cache is performed to produce models of different properties of the cache, including the number of entries, distribution of access times, distribution of entry sizes, etc. The models are then used to guide decisions that involve those properties. The size of the samples can be adjusted to balance the cost of acquiring the samples against the confidence level of the models produced by the samples. To achieve randomness, entries are stored using prefixes of addresses generated via a message-digest function.
1. A method of managing a cache, comprising:
in at least one computer communicatively connected to said cache,
determining a percentage of entries to be removed from said cache based on a size model indicative of entry sizes in said cache, said size model based on a statistical sampling of at least parts of said cache;
determining a cutoff time based on said percentage of entries and a last access model indicative of access times of entries in said cache, said last access model based on said statistical sampling; and
removing, from said cache, entries whose last access time is older than said cutoff time.
2. The method of claim 1 , further comprising:
based on said last access model, estimating a percentage of cache entries whose last access time is earlier than a reference point in time.
3. The method of claim 1 , further comprising:
based on said size model, estimating a cumulative size of all cache entries.
4. The method of claim 1 , wherein obtaining said statistical sampling of said cache comprises:
randomly selecting a directory of said cache;
reading entries of said randomly selected directory of said cache; and
comparing a number of said entries of said randomly selected directory of said cache to a predefined sample size.
5. The method of claim 4 , wherein said number of said entries of said randomly selected directory of said cache is smaller than said predefined sample size, further comprising:
repeating said randomly selecting step and said reading step until said number of said entries matches or exceeds said predefined sample size.
6. The method of claim 1 , further comprising:
receiving a cache entry; and
generating a cache address for said cache entry using a message-digest function.
7. The method of claim 6 , further comprising:
storing said cache entry in said directory of said cache based on prefixes of said cache address.
8. A system, comprising:
at least one processor;
at least one non-transitory computer-readable storage medium accessible by said at least one processor and carrying computer instructions executable by said at least one processor, wherein when executed by said at least one processor said computer instructions are operable to perform:
determining a percentage of entries to be removed from a filesystem-based cache based on a size model indicative of entry sizes in said cache, said size model based on a statistical sampling of said cache;
determining a cutoff time based on said percentage of said entries to be removed from said cache and a last access model indicative of access times of entries in said cache, said last access model based on said statistical sampling; and
removing from said cache entries whose last access time is older than said cutoff time.
9. The system of claim 8 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
based on said last access model, estimating a percentage of cache entries whose last access time is earlier than a reference point in time.
10. The system of claim 8 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
based on said size model, estimating a cumulative size of all cache entries.
11. The system of claim 8 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
randomly selecting a directory of said cache;
reading entries of said randomly selected directory of said cache;
comparing a number of said entries of said randomly selected directory of said cache to a predefined sample size; and
repeating said randomly selecting step and said reading step until said number of said entries matches or exceeds said predefined sample size.
12. The system of claim 8 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
receiving a cache entry;
generating a cache address for said cache entry using a message-digest function; and
storing said cache entry in said directory of said cache based on prefixes of said cache address.
13. A computer program product comprising at least one non-transitory computer-readable storage medium accessible by a processor and carrying computer instructions executable by said processor, wherein when executed by said processor said computer instructions are operable to perform:
determining a percentage of entries to be removed from a filesystem-based cache based on a size model indicative of entry sizes in said cache, said size model based on a statistical sampling of at least parts of said cache;
determining a cutoff time based on said percentage of said entries to be removed from said cache and a last access model indicative of access times of entries in said cache, said last access model based on said statistical sampling; and
removing from said cache entries whose last access time is older than said cutoff time.
14. The computer program product of claim 13 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
based on said last access model, estimating a percentage of cache entries whose last access time is earlier than a reference point in time.
15. The computer program product of claim 13 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
based on said size model, estimating a cumulative size of all cache entries.
16. The computer program product of claim 13 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
randomly selecting a directory of said cache;
reading entries of said randomly selected directory of said cache;
comparing a number of said entries of said randomly selected directory of said cache to a predefined sample size; and
repeating said randomly selecting step and said reading step until said number of said entries matches or exceeds said predefined sample size.
17. The computer program product of claim 13 , wherein, when executed by said at least one processor, said computer instructions are further operable to perform:
receiving a cache entry;
generating a cache address for said cache entry using a message-digest function; and
storing said cache entry in said directory of said cache based on prefixes of said cache address.
18. A method of managing a cache, comprising:
at a computer,
determining a percentage of entries to be removed from said cache based on a size model indicative of entry sizes in said cache, said size model based on a statistical sampling of at least parts of said cache; and
removing from said cache, entries whose last access time is older than a cutoff time, based on said percentage of entries.
19. The method of claim 18 , wherein the cutoff time is based on a last access model of access times of entries in said cache.
20. The method of claim 19 , wherein the last access model is based on said statistical sampling.