Refactor pgbench results output
authorMark Wong <mark@2ndQuadrant.com>
Sat, 10 Feb 2018 01:59:44 +0000 (17:59 -0800)
committerMark Wong <mark@2ndQuadrant.com>
Sat, 10 Feb 2018 01:59:44 +0000 (17:59 -0800)
commit4d397f905511a0076841c0546173310c32f1d057
tree82546fd896403e18ba8224d8c9f65fc7c395e52a
parent1f13bc07dabb555f5f8f5680c7e59cb74ef0639b
Refactor pgbench results output

The idea is that one results document represents a test of one
particular commit.  Then it contains the results of all the tests run
against that commit.  I think it makes more sense to restructure the
pgbench results to be like:

{
    "pgbench": {
        "ro": {
            "results at a scale factor": {
                "number of clients": {
                    "std": 4389.727756305762,
                    "metric": 41390.590287,
                    "median": 43137.716637,
                    "results": [
                        {
                            ... pgbench results of a test ...
                        }
                        ...
                    ]
                },
                ....
        },
        "rw": {
            ... read write test results ...
    },
    ... other tests results
}
client/benchmarks/pgbench.py