mwhittaker / frankenpaxos

A collection of state machine replication protocols

Home Page:https://mwhittaker.github.io/frankenpaxos/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

FrankenPaxos

This directory contains Scala implementations of various state machine replication protocols including Compartmentalized MultiPaxos, Matchmaker Paxos, Bipartisan Paxos, MultiPaxos, CASPaxos, CRAQ, EPaxos, Fast Paxos, and Mencius.

Getting Started

Make sure you have Java, Scala, and sbt installed. We tested everything with Java version 1.8.0_131 and with sbt version 1.1.6 on Ubuntu 18.04. You can install these however you like (e.g., we use this script and this script).

You can build and run all of frankenpaxos' code using sbt. All the code in this repository can be compiled to bytecode that you can run on the JVM and can also be compiled to Javascript that you can run in the browser. As a result, the project is split into two subprojects: frankenpaxosJVM for the code that compiles to bytecode and frankenpaxosJS for the code that compiles to Javascript. frankenpaxos is a parent project of the two.

$ sbt
sbt:frankenpaxos> frankenpaxosJVM/compile  # Build the JVM code.
sbt:frankenpaxos> frankenpaxosJS/compile   # Build the Javascript code.
sbt:frankenpaxos> frankenpaxos/compile     # Build all the code.
sbt:frankenpaxos> frankenpaxosJVM/test     # Run the tests.
sbt:frankenpaxos> frankenpaxosJS/fastOptJS # Compile to Javascript.
sbt:frankenpaxos> frankenpaxosJVM/assembly # Assemble a JAR.

$ # Run the JAR. Replace <main> with one of the executable classes. You can use
$ # tab completion to see the full list of possibilities.
$ java -cp jvm/target/scala-2.12/frankenpaxos-assembly-0.1.0-SNAPSHOT.jar <main>
$
$ # For example, we can run the echo server.
$ java -cp jvm/target/scala-2.12/frankenpaxos-assembly-0.1.0-SNAPSHOT.jar \
>     frankenpaxos.echo.ServerMain --help

Core code is in the shared directory, JVM-specific code is in the jvm/ directory, and Javascript-specific code is in the js/ directory.

Running Benchmarks

See the benchmarks/ directory for information on running benchmarks.

Running in the Browser

The code in this repository is compiled to Javascript using Scala.js. With a bit of HTML and Javascript, we can visualize the execution of a distributed system in the browser. For example, visit https://mwhittaker.github.io/frankenpaxos/js/src/main/js/echo/echo.html for a visualization of a simple echo protocol.

To run the visualization locally, run sbt frankenpaxosJS/fastOptJs and then run an HTTP server on port 8000 in the root frankenpaxos directory (e.g., by running python -m http.server 8000). Open http://localhost:8000/js/src/main/js/echo/echo.html. The source code for this example can be found in the following files:

Using Eclim

Eclim is a vim plugin that lets you use all the features and functionality of Eclipse from within vim. To use eclim with this project, first install eclim and all necessary dependencies. Then, do the following:

  1. Run eclipse from within sbt. This will create a .project file and a .classpath file in the jvm/ directory.

  2. Copy jvm/.project and jvm/.classpath into the root directory.

  3. Remove the linkedResources section in .project. It is not needed.

  4. Remove the kind="src" class path entries from the top of .classpath. Replace them with the following:

    <classpathentry kind="src" path="shared/src/main/scala"/>
    <classpathentry kind="src" path="shared/src/test/scala"/>
    <classpathentry kind="src" path="js/src/main/scala"/>
    <classpathentry kind="src" path="jvm/src/main/scala"/>
    <classpathentry kind="src" path="jvm/target/scala-2.12/src_managed/main"/>
    
  5. Open up vim and run :ProjectCreate . -n scala or :ProjectRefresh if the project already exists.

Updating Github Pages

git checkout master
git branch -f gh-pages master
git checkout gh-pages
git add -f js/target/scala-2.12/frankenpaxos-fastopt.js*
git commit
git push -f origin gh-pages

About

A collection of state machine replication protocols

https://mwhittaker.github.io/frankenpaxos/


Languages

Language:Scala 63.0%Language:Python 26.8%Language:JavaScript 8.2%Language:HTML 1.4%Language:Shell 0.5%Language:CSS 0.1%