0

Azure provides monitor to the incoming request to the Cosmos. When I am alone working on my Cosmos DB, ran a simple select vertex statement(eg., g.V('id')). Then I monitored the incoming request, it shows around 10. But for sure I know i'm the only person accessed. I also tried traversing through the graph in a single select query the Request count is huge (around 100).

Do anybody noticed the metrics? We are assuming the request code is huge for an hour in production cause the performance slowness. Is the metric is trustworthy to believe or how to find the incoming request to the cosmos?

James Z
  • 12,209
  • 10
  • 24
  • 44
Sabarish
  • 59
  • 1
  • 6
  • Not sure I understand the question. Are you talking about RU cost? If so, that's tied to the operation itself, not the request load from different users. Each operation has a specific cost. But without knowing how your collection is set up (e.g. partitioning), or exact queries you're running, there's really no way to know why your operations have the RU costs they do. Graph query cost will vary quite a bit depending on the direction of the traversal, combined with partitioning strategy. – David Makogon Nov 22 '19 at 12:06
  • @DavidMakogon thanks for your comment. I am not talking about RU cost, number of incoming request, you can check in the overview or monitor. – Sabarish Nov 22 '19 at 13:18
  • Ok - I would edit your question for clarity. But also, 100 requests, as mentioned in your original question, isn't really "huge." And there's no way to guess what's causing your requests, since you haven't provided any code. – David Makogon Nov 22 '19 at 13:35
  • @DavidMakogon I haven't tried with code, we have an options in Azure portal to check this. – Sabarish Nov 27 '19 at 06:53
  • @Sabarish The metrics will report the underling storage operations done as opposed to the Gremlin requests issued. A single gremlin request may perform one or more storage operations, depending on the nature of the traversal being executed. – Oliver Towers Jan 21 '20 at 21:15
  • Thanks @OliverTowers, yes I do think the same. :) – Sabarish Jan 23 '20 at 16:31

0 Answers0