Alert conditions and options

[almog-orca]

Problem Statement

The scenario:
We see that when we use alerts that should assist with identifying long-term [or short term] degradations in UI facing endpoints, like "transaction time increase overtime" where we would like to get alerted if there's overtime [over a month or two] increase of the transaction time, we are seeing that We can only define compared to "1week/1month" ago, but the comparsion is done on the minute/hour level - where Due to the nature of our product/service, things shifts up and down in different days and we get many FPs;

For example - usually on Sunday's we have less traffic, so comparsions on Sundays will always have huge diffs in a matter of minutes/hours aggragations - see attach picture for better clarification.

Question/request:
Is there a way to provide a bit more granularity around alerts, for “longer” and “shorter” period-of-time alerts?
For example, I would like to alert and compare, once every 24h~, the -AVG of P90- 1/2/3/4 week/s ago, and compare it to -AVG of P90 of the current week' [or day/s / month/s - instead of week/s]

p.s. The same behavior is observed when configuring a "fail rate" alert/view, if we have 1 request over 5m, and 10 requests over 6 minutes, during the 1st 5 minutes [with the single request] if request fails, we get 100% fail rate for those 5 minutes, and which is making alerting hard to be FP tolerant

Solution Brainstorm

No response

Product Area

Alerts

[getsantry]

Assigning to @getsentry/support for routing ⏲️