Query engine for data points

[vonbirdie]

• Filtering
  • 1 metric required
  • Temporal
  • Between two fixed timestamps
  • Relative from now
  • Tags
  • At least one tag required?
    • Wildcard on
    • tagv
    • tagk?
    • "Or" on
    • tagv
    • tagk?
    • "And" on
    • tagv
    • tagk?
    • "Not" on
    • tagv
    • tagk?
• Arithmetic?
  1. Rate calculation
  2. Interpolation
  3. Aggregation on filter variables
  4. Aggregation on time (downsampling)

Other query implementations are InfluxDB and Prometheus.

[vonbirdie]

Possible query syntax?

SELECT RATE(system.cpu.utilization{host=*,cpu=0|1,dc!=hk})
    GROUP BY TAG KEY avg(host)
    GROUP BY time(10m);

Different alternatives for time bounds:

BETWEEN 12312312321 AND 2132131312321
SINCE 123123123123
WHERE time > now() - 3h

[vonbirdie]

Another proposal for specifying the time:

BETWEEN [time] AND [time]

where time = ( [digit]m [digit]d [digit]H [digit]M [digit]s ago | now | timestamp)