Aggregation Service: GROUP BY support

[JonathanAquino-NextRoll]

trafficstars

We often need to report on a high cardinality attribute like campaign id. For ease of reporting and smaller query strings, it would be great if we could use something like GROUP BY to retrieve all of the different values of the attribute, instead of having to specify each value explicitly.

Something like:

select campaign_id, sum(impressions), sum(spend)
where date >= 2021-09-01 and date < 2021-10-01
group by campaign_id

Grouping by multiple attributes:

select campaign_id, foo, sum(impressions), sum(spend)
where date >= 2021-09-01 and date < 2021-10-01
group by campaign_id, foo

[csharrison]

Thanks this is something we are considering, but we need to weigh it with the increase in complexity for the servers. The current design of a fixed histogram allows a lot of flexibility with how the computation can be performed.

[JonathanAquino-NextRoll]

Thank you - hope this can get implemented.