SiriDB supports multiple build-in aggregation and filter functions. Using these functions can be useful to reduce network traffic.
List of the supported aggregation and filter functions:
It’s possible to select multiple aggregate functions in one query. This has some advantages over performing multiple queries since the database in this case only needs to search for the series and points once. To find the requested aggregate in the result we must add a prefix and/or suffix to the series name to make the name unique.
A prefix and/or suffix is only required when querying multiple aggregates.
# Select both the min and max grouped by 5 minutes from "series-001" select min(5m) prefix "min-", max (5m) prefix "max-" from "series-001"
The aggregate functions can be used together by parsing the result of one function to the next. It’s also possible to use the same function twice which can be useful with for example difference.
# Select the median grouped by 1 minute and return the difference for that result select median (1m) => difference () from "series-001"
Most aggregation functions accept an optional
ts (time span) argument; which determines the time window over which a set of values is grouped to a single summary value. A time span of
1w for example will create ‘buckets’ with a length of 1 week that are summarized to one value at the end of the bucket. Be aware that the start is always at 1970-01-01. So if you have a time series between 2020-04-06 and 2020-04-21, a
1w will result in 3 summary values with a timestamp at: 2020-04-09, 2020-04-16 and 2020-04-23. So the first bucket does not end a week from April 6th, as you might expect.
ts can also be a timestamp where you can use
now for example. This determines the end of the first bucket. The beginning starts at 1970-01-01. If there are points beyond that timestamp, they will be captured in subsequent buckets.
When the time span is not provided, SiriDB will usually return the summarized value at the last timestamp in the result. One exception is the
first() function which will return the first timestamp instead.
# Select the last time-stamp and the average over all values. select mean() from `my_series` # Select the first time-stamp and first value: select first() from `my_series`