Benchmarking the Scalability and Elasticity of the Riak NoSQL DBMS

Introduction

The [RELEASE project] (http://www.release-project.eu/) aims to improve the scalability of [the Erlang programming language] (https://www.erlang.org/). Scalable applications require scalable and available persistent storage to save and retrieve their business data. I evaluate the suitability of four popular NoSQL DBMSs ([Mnesia] (http://www.erlang.org/doc/man/mnesia.html), [CouchDB] (http://couchdb.apache.org/), [Riak] (http://basho.com/riak/), and [Cassandra] (http://cassandra.apache.org/)) for large-scale Erlang applications. I investigate the scalability and elasticity of the Riak NoSQL DBMS by employing the [Basho Bench] (http://docs.basho.com/riak/latest/ops/building/benchmarking/) on the [Kalkyl cluster] (http://www.uppmax.uu.se/resources/systems/) up to 100 nodes. More details about the benchmark are available in the [report.pdf] (https://github.com/amirghaffari/benchmark_riak/blob/master/report.pdf) file. The followings explain how to customize the scripts to match your specific environment.

Customization

In the cluster each node can be either a traffic generator or a Riak node. A traffic generator runs one copy of the Basho Bench that generates and sends commands to the Riak nodes. A Riak node contains a complete and independent copy of the Riak package which is identified by an IP address and a port number.

The followings are configurable in the run.sh file:

• Benchmark duration in minutes
• The ratio of traffic generator nodes to Riak nodes
• The Riak version
• Number of times that an experiment repeats
• Number of nodes that will leave and join back during the elastisity benchmark

The followings are configurable in the experiment.sh file:

• The storage backend for Riak, i.e. Bitcask or eLevelDB
• Erlang path for compiling Riak and Basho Bench
• R statistics language path which is needed for generating graphs
• The name and IP address of the nodes in the cluster

There are two general config files for Basho Bench in the basho_bench_config directory, one for the HTTP interace and the other for the Protocol Buffers interface. The followings are configurable in the config files:

• The number of worker process on each traffic generator node
• Database commands (e.g. get, insert, update) used in the benchmark

How to run

$ git clone git://github.com/amirghaffari/benchmark_riak
$ cd benchmark_riak
After customizing the `run.sh` and `experiment.sh` files:
$ ./run.sh