Currently, there doesn't exist a great way for us to "fail" a test due to the results of a Google benchmark. Today, we just run the benchmarks and then send the data up into Cedar to be tracked as a regular performance result.

We want a way to be able to set a threshold limit on the result of a Google benchmark test. If the benchmark reports a result for a metric that is above one of the thresholds we set, the test should fail and that result should be reported to the engineer that ran the test.