Performance of Flux vs InfluxQL

, ,
1

I have used InfluxDB for almost a year now, I expected after releasing stable version 2.0 Flux performance problem should be resolved! Can anyone tell me why InfluxQL outperforms in the order of 100 for a simple getting raw filtered data? Is there anything that I miss? In the below, I show two simple identical queries first is with Flux which took 30 seconds and the other is InfluxQL and took 0.3 seconds.
Flux:
from(bucket: “kpi”)
|> range(start: 2020-11-11T00:00:00.000Z, stop: 2020-11-11T23:59:00.000Z)
|> filter(fn: ® => r["_measurement"] == “mem” and r["_field"] == “cpu”)
|> yield(name: “raw”)
InfluxQL:
SELECT mem from mem where time >= ‘2020-11-11T00:00:00.000Z’ and time < ‘2020-11-11T23:59:00.000Z’ GROUP BY *

The Flux query took 32 Seconds and the Influx took 0.5. I am using InfluxDB 2.0.2

1 Like
2

I too have noticed significant performance regressions with Flux and InfluxDB OSS 2 GA releases when compared to like queries in InfluxQL, or indeed with the same Flux queries running on the InfluxDB OSS 2 Beta versions before the storage engine was reverted to the 1.8 version.

What is the roadmap to optimise Flux on InfluxDB OSS?

Thanks,
Graham