Optimize ASTPkg.Analyze

[williamhbaker]

In ASTPkg.Analyze, calling C.flux_analyze(...) takes on the order of 5-8ms for every query on my machine (2019 MacBook Pro, 2.6 GHz 6-core i7) running on Influxdb OSS v2.0.7.

This isn't a very big deal for long-running queries in the 100's or 1000's of ms (or longer), but for fast-running queries it can be a large contributor to the total response time of the query.

For example, a flux query like the following executed on a fairly small set of data (generated from the influxdb-comparisons tool using the devops use case) takes about 10ms to run for me locally, with about 6ms coming from the C.flux_analyze(...) call:

from(bucket: "benchmark_db") 
  |> range(start: -5m) 
  |> filter(fn:(r) => r._measurement == "cpu" and r._field == "usage_user") 

An equivalent InfluxQL query like the following will complete in ~1ms, so the flux query is about 10x slower, with a large amount of that extra time coming from the aforementioned method:

SELECT usage_user from cpu where time >= '<now-5 minutes>' and time < '<now>'

[github-actions]

This issue has had no recent activity and will be closed soon.