09 Aug 2023 11:01 AM
Hi Team,
I have a question for you guys if you could share to me knowledge how exactly SLO calculates in case if we have several services in one SLO. From my practice, I always choose only one service to avoid any problem, but in this case a question from colleagues who want choose several.
For example - Evaluating period -1h and I have three services and want Service-level availability SLO for them:
Service A ( 1 failed requests / 100 request/h ) SLO for A - 99%
Service B ( 0 failed requests / 1 request/h ) SLO for B - 100%
Service C ( 500 failed requests / 1 000 requests/h ) SLO for C - 50%
Service A,B,C (management zone filter) I have SLO for ABC - some_percentage
For example - 80.363%. I can see it in wizard at Evaluate step.
How exactly calculation takes place?
From real tests I have two services with 100%, and 90%, and I can see that SLO for management zone for these two services not just (100 + 90)/2 = 95% 😃
I would appreciate any help with this 🙂
Regards,
Alex Romanenkov
09 Aug 2023 07:42 PM
Hi Alex,
Isnt it basically: all the requests compared to all the succeeded requests?
So if you have 100% and 95% for each service, it can be different. Since it can happen that the 95 procent service has more requests in total?
SLO example could be:
If you use 2 services, the requests of both are combined to do the calculation. You can replay it by just fetching the metrics in data explorer en do the calculation yourself 🙂
KR.
Michiel