Azure Analysis Services Memory Dashboard & Cache

Question

I'm trying to understand what the memory metric in Azure portal means for AAS.

~6:30pm I got an error (see Image 1) saying 'Memory error: You have reached the maximum allowable memory allocation for your tier. Consider upgrading to a tier with more available memory.' I then scaled up from S0 (10 GB) to S1 (25GB) ~10pmEST.

My questions are:

• Why does the memory spike up during processing?
• Why does the memory settle back down at a baseline of ~5.5GB?

IMAGE 1 - Error
![140074-2021-10-12-22h10-04.png][1]

IMAGE 2 - Memory Dashboard in Azure Portal

I've found these links helpful:

• Memory Caching Issue in AAS
• AAS Memory & Compression
• AAS Memory Utilization
• AAS Compression [1]: /api/attachments/140074-2021-10-12-22h10-04.png?platform=QnA 3: https://learn.microsoft.com/en-us/answers/questions/348937/memory-caching-issue-in-azure-analysis-services.html

Answer 1

Hello @Mariah and welcome to Microsoft Q&A.

When AAS does processing, it reads all the data into memory, does some work on it, and then afterwards writes the data back. This is the spike, holding all the data in memory and working on it.

The "at rest" memory level, is the cache AAS holds to answer queries.

It is like a gasoline-powered automobile. You turn it on, and the engine is running idle. Then when you accelerate, the engine works harder. When you stop accelerating, it returns to idle. Even at idle, the engine is running. Only when turn the car off does engine stop.

Does this make sense?