CacheToolsUtils 4.2

Cachetools Utilities

cachetools-utils

Classes to add key prefix and stats to cachetools classes and use redis and memcached as storage backends, and other cache-related utils.

Thoughts about Caching

Caching is a key component of any significant Web or REST backend so as to avoid performance issues when accessing the storage tier, in term of latency, throughput and resource usage.

• Shared Cache

  A convenient setup is to have one shared cache storage tier at the application level, which is accessed through wrappers to avoid collisions between cache functions, basically by prepending keys with some prefix.

  Depending on the access pattern, it may or may not be useful to put a multiple-level caching strategy in place, with a local in-process cache and a higher-level inter-process and inter-host cache such as Redis or MemCached.

  When using a global shared cache, it should be clear that the cache may hold sensitive data and its manipulation may allow to change the behavior of the application, including working around security by tampering with the application authentication and authorization guards.

• Latency

  In order to reduce latency, as most time should be spent in network accesses, reducing the number of trips is a key strategy. This suggests combining data transfers where possible through higher-level interfaces and queries, both at the HTTP level and at the database level.

  Denormalizing the relational data model may help. Having an application-oriented view of the model (eg JSON objects rather than attributes and tables) can help performance, at the price of losing some of the consistency warranties provided by a database. The best of both word may be achieved, to some extent, by storing JSON data into a database such as Postgres.

  Invalidating data from the cache requires a detailed knowledge of internal cache operations and are not very easy to manage at the application level, so devops should want to avoid this path if possible, possibly by relying on a time-based cache expiration aka TTL (time-to-live).

• Throughput

  Write operations need to be sent to storage. Depending on transaction requirements, i.e. whether some rare data loss is admissible, various strategies can be applied, such as updating in parallel the cache and the final storage. Yet again, this strategy requires a deep knowledge of the underlying cache implementation, thus is best avoided most of the time.

  Read operations can be cached, at the price of possibly having inconsistent data shown to users. LFU/LRU cache strategies mean that inconsistent data can be kept in cache for indefinite time, which is annoying. A TLL expiration on top of that makes such discrepancies bounded in time, so that after some time the data shown are eventually up to date.

Basically the application should aim at maximizing throughput for the available resources whilst keeping the latency under control, eg 90% of queries under some limit.

Module Documentation

This module provide the following cache wrappers suitable to use with cachetools:

• Some classes provide actual storage or API to actual storage. For this purpose a cache is basically a key-value store, aka a dictionary, possibly with some constraints on keys (type, size) and values (size, serialization).

• Other classes add features on top of these, such as using a prefix so that a storage can be shared without collisions or keeping usage and efficiency statistics.

PrefixedCache

Add a key prefix to an underlying cache to avoid key collisions.

import cachetools
import CacheToolsUtils as ctu

ct_base = cachetools.TTLCache(maxsize=1048576, ttl=600)
foo_cache = ctu.PrefixedCache(ct_base, "foo.")
bla_cache = ctu.PrefixedCache(ct_base, "bla.")

@cachetools.cached(cache=foo_cache)
def foo(…):
    return …

@cachetools.cached(cache=bla_cache)
def bla(…):
    return …

StatsCache

Keep stats, cache hit rate shown with hits().

scache = ctu.StatsCache(cache)

TwoLevelCache

Two-level cache, for instance a local in-memory cachetools cache for the first level, and a larger shared redis or memcached distributed cache for the second level. Whether such setting can bring performance benefits is an open question.

cache = ctu.TwoLevelCache(TTLCache(…), RedisCache(…))

There should be some consistency between the two level configurations so that it makes sense. For instance, having two TTL-ed stores would suggest that the secondary has a longer TTL than the primary.

MemCached

Basic wrapper, possibly with JSON key encoding thanks to the JsonSerde class. Also add a hits() method to compute the cache hit ratio with data taken from the memcached server.

import pymemcache as pmc

mc_base = pmc.Client(server="localhost", serde=ctu.JsonSerde())
cache = ctu.MemCached(mc_base)

@cachetools.cached(cache=cache)
def poc(…):

Keep in mind MemCached limitations: key size is limited to 250 bytes strings where some characters cannot be used, eg spaces, which suggest some encoding such as base64, further reducing the actual key size; value size is 1 MiB by default.

PrefixedMemCached

Wrapper with a prefix. A specific class is needed because of necessary key encoding.

pcache = ctu.PrefixedMemCached(mc_base, prefix="pic.")

RedisCache

TTL'ed Redis wrapper, default ttl is 10 minutes. Also adds a hits() method to compute the cache hit ratio with data taken from the Redis server.

import redis

rd_base = redis.Redis(host="localhost")
cache = ctu.RedisCache(rd_base, ttl=60)

Redis stores arbitrary bytes. Key and values can be up to 512 MiB. Keeping keys under 1 KiB seems reasonable.

PrefixedRedisCache

Wrapper with a prefix and a ttl. A specific class is needed because of key encoding and value serialization and deserialization.

pcache = ctu.PrefixedRedisCache(rd_base, "pac.", ttl=3600)

cacheMethods and cacheFunctions

This utility function create a prefixed cache around methods of an object or functions in the global scope. First parameter is the actual cache, second parameter is the object or scope, and finally a keyword mapping from function names to prefixes.

# add cache to obj.get_data and obj.get_some
ctu.cacheMethods(cache, obj, get_data="1.", get_some="2.")

# add cache to some_func
ctu.cacheFunctions(cache, globals(), some_func="f.")

Install

Install with pip:

pip install CacheToolsUtils

See above for example usage.

License

This code is public domain.

Versions

4.2 on 2022-08-05

Fix minor typo in a badge.

4.1 on 2022-08-05

Code reformating. Improved documentation. Improved checks. Improved Makefile.

4.0 on 2022-03-13

Remove StatsRedisCache and StatsMemCached by moving the hits() method to RedisCache and MemCached, respectively. The two classes still exist for upward compatibility, but are deprecated. Improve test coverage, now only 4 not-covered lines. Improve documentation.

3.0 on 2022-03-06

Use simpler kwargs approach for caching methods and functions. Add a gen parameter for caching methods and functions. Improve documentation.

2.0 on 2022-02-24

Add cacheMethods and cacheFunctions. Improve documentation. 100% coverage test.

1.1.0 on 2022-01-30

Improve documentation. Add TwoLevelCache. Add 100% coverage test.

1.0.0 on 2022-01-29

Add set, get and delete forwarding to RedisCache, so that redis classes can be stacked.

0.9.0 on 2022-01-29

Initial version extracted from another project.

TODO

• add a close?
• rename hits hit_rate?
• add other efficiency statistics?