Memtier-Benchmark Director

This is a front-end tool that makes it easier to run successive benchmarking sessions using the memtier_benchmark tool by automating tasks such as:

• Setting up and tearing down Redis instances
• Running memtier_benchmark
• Collecting and organizing all data, logs, etc.
• Providing a simple web front to access data and a visual summary

Getting Started

Clone this repo and cd into it. Next, set up a local Python virtualenv and install:

mkdir .env
virtualenv .env
. .env/bin/activate
python setup.py install

To run, create a benchmark JSON file and run:

mbdirector benchmark -s mybenchmark.json

This will set up the necessary targets, execute benchmarks, and store the results in a results directory. To access the results run mbdirector serve from the same directory and point your browser to http://localhost:8080.

Benchmark Configuration

The benchmark configuration defines:

1. One or more target. A target is a Redis database which is set up for the purpose of running a benchmark and teared down when it is complete.

   Currently the only supported target type is a local Redis process. A useful "TODO" item would be to support Redis Enterprise API to provision a database of arbitrary configuration.

2. One or more benchmark. A benchmark is a description of how to run memtier_benchmark (i.e. with specific arguments controlling size of elements, pipelining, etc.).

When executed, mbdirector runs a combination of all configured benchmarks and all configured targets.