Skip to main content

Consistent hashing for Python

Project description

Copyright (c) 2010 by Joachim Bauch, mail@joachim-bauch.de http://www.joachim-bauch.de/projects/python-continuum/

python-continuum provides a function to do consistent hashing. See wikipedia [1] for more informations about this technique.

First we need the continuum object that serves as storage for server informations and can later be used to resolve keys:

>>> from continuum import Continuum
>>> c = Continuum()

Empty continuums obviously can’t be queried:

>>> c.resolve('my-key1')
Traceback (most recent call last):
...
IndexError: empty continuum

Add the server objects that are available as backends:

>>> c.add_server('192.168.0.1', 8080)
<Server "192.168.0.1:8080", capacity=1>
>>> c.add_server('192.168.0.2', 8080)
<Server "192.168.0.2:8080", capacity=1>

You can also specify different capacities of the server to priorize them (this defaults to 1):

>>> c.add_server('192.168.0.3', 8080, 2)
<Server "192.168.0.3:8080", capacity=2>

Server objects can also be removed:

>>> server = c.add_server('192.168.0.4', 8080)
>>> c.remove_server(server)
>>> len(c)
3

Please note that a server can only be added once:

>>> c.add_server('192.168.0.2', 8080)
Traceback (most recent call last):
...
TypeError: server already added

After all servers have been added, the continuum can be queried for the server that should be used for a given (string) key:

>>> c.resolve('my-first-key')
<Server "192.168.0.1:8080", capacity=1>
>>> c.resolve('my-other-key')
<Server "192.168.0.3:8080", capacity=2>

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page