Automated Benchmarks

[marceloboeira]

The current benchmark structure is not quite great.

See https://github.com/14-bits/voik/blob/master/src/benchmark.rs for more info.

Acceptance Criteria

The benchmark must run on every build, and it must inform a reasonable metric of MB/s from reads and writes of a single commit-log shard. The side of the index and the log file should be configurable, as much as the size and total amount of records.

e.g.: "run a test with 10MB index, 20MB log, 1M messages of 100kb each"...

[jaywalker76]

Hi @marceloboeira, besides having a look at the consensus protocols, I'd like to work on this issue, if that's ok. Please note that I don't have experience with Rust and as such my turnover rate may not be the best. If you're ok with that I can take over this issue.

[marceloboeira]

@jaywalker76 feel free to look into it. I'll be out for a couple days and won't be able to work too much on it.

You can write the tests based on the commit_log sub-crate at the moment, which is the most important part. Test for read and write throughput for the "lib" --> https://doc.rust-lang.org/1.6.0/book/benchmark-tests.html