Handling large order/orderlines scenarios

Question

We are currently investigating our options regarding the storage options for a new application. Cosmos DB is on our radar and I stumbled upon these docs regarding hierarchical data.

The example uses the order/orderdetails scenario and that model closely relates to our model. But I have some questions about whether cosmos db is still a good fit if the number of orderlines is very large, say 10000 lines. How would that work in terms of updating a select set of orderlines, would we need to read the whole document in memory and how cost effective would such scenario be?

Answer 1

Hi @Peter Bons , welcome to Microsoft Q&A forum.

Azure Cosmos DB gives us the advantage of having the entire relationship defined in single document as de-normalized data that means embedding all the information into single document which would help us in retrieving everything in single read operation. Now this looks good when we have one-to-few relationship between entities. But if there are lot of entities or ever growing entities then we need to think of alternatives. Reason being retrieving a big document could affect our RUs consumption and thus impacting our cost.

In this case its better to form the weak relationship between documents as mentioned in the below article:

Data modeling in Azure Cosmos DB

This article provides a very nice explanation of multiple scenarios that we could encounter based on Cosmos DB design. I would suggest you to go through it once.

Also how to form the weak links and query them can be referred in below thread:

https://learn.microsoft.com/en-us/answers/questions/571404/index.html

Please let us know if this helps or else we can discuss further on the same.

----------

Please don't forgot to click on accept it as answer button wherever the information provided helps you. This can be beneficial to other community members as well.