quantize Metrics Operator
You can use the quantize
operator to control the Sumo’s quantization behavior, which is described in detail in Metric Quantization.
You can specify:
- The size of the time buckets across which Sumo aggregates your metrics. If you do not specify a quantization interval, Sumo determines an optimum size for time buckets, as described in Automatic quantization at query time.
- The rollup type that Sumo uses to aggregate the individual data points in a time bucket, which can be one of
avg, min, max, sum,
orcount
. If you do not specify a rollup type in thequantize
clause of your query, for each time bucket, Sumo presents the average of the data points in that bucket.
quantize syntax
The quantize
operator must appear immediatdly after your query selector, before any other operators.
metrics query | quantize to INTERVAL [using ROLLUP] [drop last]
where:
INTERVAL
is the duration over which you want to quantize the metrics, in seconds (s
) , minutes (m
), hours (h
), or days (d
).ROLLUP
isavg, min, max, sum
, orcount
.drop last
causes the last time bucket to be dropped, if the end of that bucket is after the end of the query time range.
In the Metrics Explorer, you must switch to Advanced Mode to enter the drop last
option.
quantize examples
Set time bucket size
The quantize
clause in this metric query sets the time bucket size to 5 minutes. Sumo will aggregate the metrics in each time bucket using the default rollup type, avg
.
_sourceCategory=hostmetrics | quantize to 5m
Set time bucket size and rollup type
The quantize
clause in this metric query sets the time bucket size to 10 minutes, and specifies the sum
rollup type. Sumo will sum the metric values in each 10 minute time bucket and return that value.
metric=CPU_User cluster=kafka | quantize to 10m using sum
Set time bucket size, rollup type, and drop last
The quantize
clause in this metric query sets the time bucket size to 10 minutes, and specifies the sum
rollup type. Sumo will sum the metric values in each 10 minute time bucket and return that value. If the last time bucket ends after the end of the query time range, that bucket is dropped.
metric=CPU_User cluster=kafka | quantize to 10m using sum drop last