Tom Cocagne <[email protected]>
v2.0, January 2013
This repository contains the results of my first foray into providing both a useful and educational Paxos implementation. The repository is in pretty good shape and still fills its intended role but the design includes explicit support for sending network messages. I now consider this something of an anti-pattern for Paxos implementations and recommend using an even more tightly constrained library that implements the core algorithm and nothing else, including messaging. To that end, I've put together the python-composable-paxos repository and a multi-paxos-example based upon it. Those repositories should generally be preferred over this one. Additionally, I've written an introductory paper on Paxos and Multi-Paxos that may be useful for individuals like myself that find the existing papers somewhat opaque.
Essential Paxos provides basic implementations of the Paxos algorithm. The distinguishing characteristic of this implementation, as compared to other freely available and open-source implementations, is that this library is independent of application domains and networking infrastructures. Whereas most Paxos implementations are deeply and inextricably embedded within application-specific logic, this implementation focuses on encapsulating the Paxos algorithm within opaque and easily re-usable classes.
This library provides an algorithmically correct Paxos implementation that may be used for educational purposes in addition to direct use in networked applications. This implementation is specifically designed to facilitate understanding of both the essential Paxos algorithm as well as the practical considerations that must be taken into account for real-world use.
Implementations in both Python and Java are provided.
$ easy_install essential-paxosThis module provides a direct and minimal implementation of the essential Paxos algorithm. The primary purpose of this module is educational as it allows for easy contrast between the implementation of the pure algorithm and that of the one enhanced for practicality.
This module enhances the essential Paxos algorithm and adds support for such things as leadership tracking, NACKs, and state persistence.
This module provides a fully-functional Paxos implementation that employs a simple heartbeating mechanism to detect leadership failure and initiate recovery.
This module provides an enhanced version of practical.py that supports the use of external failure detectors to drive leadership management. This module does not provide a fully-functional solution to leadership management, as does functional.py. However, it may serve as the basis for much more flexible, application-specific leadership management.
Correct implementations of the Paxos algorithm require saving Acceptor state to persistent media prior to sending Promise and Accepted messages over the network. This is necessary to ensure that promises made to external entities will never be reneged upon should the application crash and recover at an inopportune time. This module implements a very simple mechanism for efficiently saving application state to disk.
The Java implementation is functionally identical to that of the Python implementation and is broken out into Java packages that mirror the python modules. Refer to the corresponding Python APIs for class and method documentation.
The test directory of the root source code repository contains the unittest
files used to excersise the implementation. The primary tests are written in
Python and correspond to the essential, practical, functional, and durable
modules. The Java tests, which are also written in Python, wrap the Java classes
with a compatible interface and use the Python unit tests to exercise the Java
implementation. The Jython interpreter is required for running these tests but
it is not required at runtime.