Opt-In for improved aggregate performance

Question

In the August 30th Azure Friday video, it was mentioned that there has been work to improve aggregate performance in CosmosDB.  It was also stated that this update was available by request only for now.  How do we make this request for one of our CosmosDB databases?

Thank you!

Jim Hart

Answer 1

Hi 

We are checking this with Cosmos DB Product Group. We will get back to you as soon as we have a reply.

Thanks
Navtej S

Answer 2

Hi, any update to this?

Answer 3

I am looking for an answer on this too.

Answer 4

Hi 

We have received the reply from our PG team:

For new accounts, the improvements will be available by default soon (likely this week - week of 14th Oct 2019).

For existing accounts, customers can reach out to Cosmos team via support request/CSS and we can migrate their accounts for them.

Hope this helps. Please get back to us for any questions.

Thanks
Navtej S

Answer 5

Hi

We are facing similar issue with data querying and aggregation on Cosmos.

We've reached out to support to increase the memory limits on aggregate pipeline and also increase the max memory limit of 40MB on normal queries. 

ticked id: 119101124001685

They said they enabled the aggregation cursor (not sure what it is, and not sure if this is the improvement that is mentioned in @NavtejSaini response)

Anyhow, it did not solve the issue. The support team is working on it.

Answer 6

Hi 

@shanmuk  - Please work with the support as you have already opened a ticket with them as your aggregate queries have issue of timing out. 

@Jim - I have already provided the method to do what was mentioned in the video and have proposed it as an answer, please mark it as a reply if it has helped you.

Thanks
Navtej S

Answer 7

I submitted a support request through the Azure portal.  I'm waiting on a response and update to a database we would like to test on.

Thank you

Answer 8

The new upgraded Mongo Server version on Cosmos resolved the issue. (Version 3.6)

MSFT Support upgraded the version on an existing Cosmos Instance that we had.