Mako is a set of tools for pre-production performance testing. It offers:

• data storage
• charting
• statistical aggregation
• automated regression analysis
• workflows for triaging potential regressions

Mako is narrowly focused on meeting the performance testing needs of Google open source projects. At this time, Mako is not intended to be used by projects outside this limited scope.

Mako client libraries are designed to be called from a client project’s performance test code. See the client project’s documentation for how to run their performance tests.

Since Mako performance tests upload data to https://mako.dev, they require access rights to run. Please see ACCESS.md.

Results from performance tests that use Mako are visible to the world. Visit https://mako.dev to browse the results. Find dashboard documentation at https://mako.dev/help.

The Mako command-line tool can be used to gain programmatic access to the same Mako data that’s available on the dashboard. This tool is used by benchmark owners to manage their benchmarks and runs.

To learn about how to use the Mako command-line tool, please read CLI.md.

If you’re writing a new performance test that will use a Mako client to store results in https://mako.dev and to guard against performance regressions, please read GUIDE.md.

There are example performance tests in the examples/ folder.

The most common programmatic use of Mako is to create a new run from performance test data, and for that use case we recommend Quickstore. But if you need different kinds of access to the data (e.g. updating or deleting existing data), you’ll want to use the C++ or Go storage library.