DiskCache_ is an Apache2 licensed disk and file backed cache library, written
in pure-Python, and compatible with Django.
The cloud-based computing of 2021 puts a premium on memory. Gigabytes of empty space is left on disks as processes vie for memory. Among these processes is Memcached (and sometimes Redis) which is used as a cache. Wouldn't it be nice to leverage empty disk space for caching?
Django is Python's most popular web framework and ships with several caching backends. Unfortunately the file-based cache in Django is essentially broken. The culling method is random and large caches repeatedly scan a cache directory which slows linearly with growth. Can you really allow it to take sixty milliseconds to store a key in a cache with a thousand items?
In Python, we can do better. And we can do it in pure-Python!
In : import pylibmc In : client = pylibmc.Client(['127.0.0.1'], binary=True) In : client[b'key'] = b'value' In : %timeit client[b'key']
10000 loops, best of 3: 25.4 µs per loop
In : import diskcache as dc In : cache = dc.Cache('tmp') In : cache[b'key'] = b'value' In : %timeit cache[b'key']
100000 loops, best of 3: 11.8 µs per loop
Note: Micro-benchmarks have their place but are not a substitute for real measurements. DiskCache offers cache benchmarks to defend its performance claims. Micro-optimizations are avoided but your mileage may vary.
DiskCache efficiently makes gigabytes of storage space available for caching. By leveraging rock-solid database libraries and memory-mapped files, cache performance can match and exceed industry-standard solutions. There's no need for a C compiler or running another process. Performance is a feature and testing has 100% coverage with unit tests and hours of stress.
Daren Hasenkamp_, Founder --
"It's a useful, simple API, just like I love about Redis. It has reduced the amount of queries hitting my Elasticsearch cluster by over 25% for a website that gets over a million users/day (100+ hits/second)."
Mathias Petermann_, Senior Linux System Engineer --
"I implemented it into a wrapper for our Ansible lookup modules and we were able to speed up some Ansible runs by almost 3 times. DiskCache is saving us a ton of time."
Does your company or website use
DiskCache? Send us a
message <firstname.lastname@example.org> and let us know.
Daren Hasenkamp: https://www.linkedin.com/in/daren-hasenkamp-93006438/
Mathias Petermann: https://www.linkedin.com/in/mathias-petermann-a8aa273b/
DiskCache is simple with
$ pip install diskcache
You can access documentation in the interpreter with Python's built-in help function::
import diskcache help(diskcache) # doctest: +SKIP
The core of
DiskCache is three data types intended for caching.
objects manage a SQLite database and filesystem directory to store key and
FanoutCache provides a sharding layer to utilize multiple
DjangoCache integrates that with
from diskcache import Cache, FanoutCache, DjangoCache help(Cache) # doctest: +SKIP help(FanoutCache) # doctest: +SKIP help(DjangoCache) # doctest: +SKIP
Built atop the caching data types, are
Index which work as a
cross-process, persistent replacements for Python's
dict. These implement the sequence and mapping container base classes::
from diskcache import Deque, Index help(Deque) # doctest: +SKIP help(Index) # doctest: +SKIP
Finally, a number of
recipes_ for cross-process synchronization are provided
using an underlying cache. Features like memoization with cache stampede
prevention, cross-process locking, and cross-process throttling are available::
from diskcache import memoize_stampede, Lock, throttle help(memoize_stampede) # doctest: +SKIP help(Lock) # doctest: +SKIP help(throttle) # doctest: +SKIP
Python's docstrings are a quick way to get started but not intended as a
replacement for the
DiskCache Tutorial and
DiskCache API Reference.
For those wanting more details, this part of the documentation describes tutorial, benchmarks, API, and development.
DiskCache Cache Benchmarks_
DiskCache DjangoCache Benchmarks_
Case Study: Web Crawler_
Case Study: Landing Page Caching_
Talk: All Things Cached - SF Python 2017 Meetup_
DiskCache API Reference_
DiskCache Tutorial: http://www.grantjenks.com/docs/diskcache/tutorial.html
DiskCache Cache Benchmarks: http://www.grantjenks.com/docs/diskcache/cache-benchmarks.html
DiskCache DjangoCache Benchmarks: http://www.grantjenks.com/docs/diskcache/djangocache-benchmarks.html
Talk: All Things Cached - SF Python 2017 Meetup: http://www.grantjenks.com/docs/diskcache/sf-python-2017-meetup-talk.html
Case Study: Web Crawler: http://www.grantjenks.com/docs/diskcache/case-study-web-crawler.html
Case Study: Landing Page Caching: http://www.grantjenks.com/docs/diskcache/case-study-landing-page-caching.html
DiskCache API Reference: http://www.grantjenks.com/docs/diskcache/api.html
DiskCache Development: http://www.grantjenks.com/docs/diskcache/development.html
Comparisons to popular projects related to
Key-Value Stores ................
DiskCache_ is mostly a simple key-value store. Feature comparisons with four
other projects are shown in the tables below.
dbm_ is part of Python's standard library and implements a generic interface to variants of the DBM database — dbm.gnu or dbm.ndbm. If none of these modules is installed, the slow-but-simple dbm.dumb is used.
shelve_ is part of Python's standard library and implements a “shelf” as a persistent, dictionary-like object. The difference with “dbm” databases is that the values can be anything that the pickle module can handle.
sqlitedict_ is a lightweight wrapper around Python's sqlite3 database with a simple, Pythonic dict-like interface and support for multi-thread access. Keys are arbitrary strings, values arbitrary pickle-able objects.
pickleDB_ is a lightweight and simple key-value store. It is built upon Python's simplejson module and was inspired by Redis. It is licensed with the BSD three-clause license.
================ ============= ========= ========= ============ ============ Feature diskcache dbm shelve sqlitedict pickleDB ================ ============= ========= ========= ============ ============ Atomic? Always Maybe Maybe Maybe No Persistent? Yes Yes Yes Yes Yes Thread-safe? Yes No No Yes No Process-safe? Yes No No Maybe No Backend? SQLite DBM DBM SQLite File Serialization? Customizable None Pickle Customizable JSON Data Types? Mapping/Deque Mapping Mapping Mapping Mapping Ordering? Insert/Sorted None None None None Eviction? LRU/LFU/more None None None None Vacuum? Automatic Maybe Maybe Manual Automatic Transactions? Yes No No Maybe No Multiprocessing? Yes No No No No Forkable? Yes No No No No Metadata? Yes No No No No ================ ============= ========= ========= ============ ============
================ ============= ========= ========= ============ ============ Project diskcache dbm shelve sqlitedict pickleDB ================ ============= ========= ========= ============ ============ Tests? Yes Yes Yes Yes Yes Coverage? Yes Yes Yes Yes No Stress? Yes No No No No CI Tests? Linux/Windows Yes Yes Linux No Python? 2/3/PyPy All All 2/3 2/3 License? Apache2 Python Python Apache2 3-Clause BSD Docs? Extensive Summary Summary Readme Summary Benchmarks? Yes No No No No Sources? GitHub GitHub GitHub GitHub GitHub Pure-Python? Yes Yes Yes Yes Yes Server? No No No No No Integrations? Django None None None None ================ ============= ========= ========= ============ ============
These are rough measurements. See
DiskCache Cache Benchmarks_ for more
================ ============= ========= ========= ============ ============ Project diskcache dbm shelve sqlitedict pickleDB ================ ============= ========= ========= ============ ============ get 25 µs 36 µs 41 µs 513 µs 92 µs set 198 µs 900 µs 928 µs 697 µs 1,020 µs delete 248 µs 740 µs 702 µs 1,717 µs 1,020 µs ================ ============= ========= ========= ============ ============
Caching Libraries .................
joblib.Memory_ provides caching functions and works by explicitly saving the inputs and outputs to files. It is designed to work with non-hashable and potentially large input and output data types such as numpy arrays.
klepto_ extends Python’s
lru_cacheto utilize different keymaps and alternate caching algorithms, such as
mru_cache. Klepto uses a simple dictionary-sytle interface for all caches and archives.
Data Structures ...............
dict_ is a mapping object that maps hashable keys to arbitrary values. Mappings are mutable objects. There is currently only one standard Python mapping type, the dictionary.
pandas_ is a Python package providing fast, flexible, and expressive data structures designed to make working with “relational” or “labeled” data both easy and intuitive.
Sorted Containers_ is an Apache2 licensed sorted collections library, written in pure-Python, and fast as C-extensions. Sorted Containers implements sorted list, sorted dictionary, and sorted set data types.
Sorted Containers: http://www.grantjenks.com/docs/sortedcontainers/
Pure-Python Databases .....................
ZODB_ supports an isomorphic interface for database operations which means there's little impact on your code to make objects persistent and there's no database mapper that partially hides the datbase.
CodernityDB_ is an open source, pure-Python, multi-platform, schema-less, NoSQL database and includes an HTTP server version, and a Python client library that aims to be 100% compatible with the embedded version.
TinyDB_ is a tiny, document oriented database optimized for your happiness. If you need a simple database with a clean API that just works without lots of configuration, TinyDB might be the right choice for you.
Object Relational Mappings (ORM) ................................
Django ORM_ provides models that are the single, definitive source of information about data and contains the essential fields and behaviors of the stored data. Generally, each model maps to a single SQL database table.
SQLAlchemy_ is the Python SQL toolkit and Object Relational Mapper that gives application developers the full power and flexibility of SQL. It provides a full suite of well known enterprise-level persistence patterns.
Peewee_ is a simple and small ORM. It has few (but expressive) concepts, making it easy to learn and intuitive to use. Peewee supports Sqlite, MySQL, and PostgreSQL with tons of extensions.
SQLObject_ is a popular Object Relational Manager for providing an object interface to your database, with tables as classes, rows as instances, and columns as attributes.
Pony ORM_ is a Python ORM with beautiful query syntax. Use Python syntax for interacting with the database. Pony translates such queries into SQL and executes them in the database in the most efficient way.
Django ORM: https://docs.djangoproject.com/en/dev/topics/db/
Pony ORM: https://ponyorm.com/
SQL Databases .............
SQLite_ is part of Python's standard library and provides a lightweight disk-based database that doesn’t require a separate server process and allows accessing the database using a nonstandard variant of the SQL query language.
MySQL_ is one of the world’s most popular open source databases and has become a leading database choice for web-based applications. MySQL includes a standardized database driver for Python platforms and development.
PostgreSQL_ is a powerful, open source object-relational database system with over 30 years of active development. Psycopg is the most popular PostgreSQL adapter for the Python programming language.
Oracle DB_ is a relational database management system (RDBMS) from the Oracle Corporation. Originally developed in 1977, Oracle DB is one of the most trusted and widely used enterprise relational database engines.
Microsoft SQL Server_ is a relational database management system developed by Microsoft. As a database server, it stores and retrieves data as requested by other software applications.
Oracle DB: https://pypi.org/project/cx_Oracle/
Microsoft SQL Server: https://pypi.org/project/pyodbc/
Other Databases ...............
Memcached_ is free and open source, high-performance, distributed memory object caching system, generic in nature, but intended for use in speeding up dynamic web applications by alleviating database load.
Redis_ is an open source, in-memory data structure store, used as a database, cache and message broker. It supports data structures such as strings, hashes, lists, sets, sorted sets with range queries, and more.
MongoDB_ is a cross-platform document-oriented database program. Classified as a NoSQL database program, MongoDB uses JSON-like documents with schema. PyMongo is the recommended way to work with MongoDB from Python.
LMDB_ is a lightning-fast, memory-mapped database. With memory-mapped files, it has the read performance of a pure in-memory database while retaining the persistence of standard disk-based databases.
BerkeleyDB_ is a software library intended to provide a high-performance embedded database for key/value data. Berkeley DB is a programmatic toolkit that provides built-in database support for desktop and server applications.
LevelDB_ is a fast key-value storage library written at Google that provides an ordered mapping from string keys to string values. Data is stored sorted by key and users can provide a custom comparison function.
DiskCache at PyPI_
DiskCache at GitHub_
DiskCache Issue Tracker_
DiskCache Documentation: http://www.grantjenks.com/docs/diskcache/
DiskCache at PyPI: https://pypi.python.org/pypi/diskcache/
DiskCache at GitHub: https://github.com/grantjenks/python-diskcache/
DiskCache Issue Tracker: https://github.com/grantjenks/python-diskcache/issues/
Copyright 2016-2022 Grant Jenks
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.