My desktop app (in development) is using an Azure SQL DB to store shared content. I want to allocate costs for storage and compute time for each client based on their usage. I think I understand how I can calcualte the storage costs, but am not sure about computing costs. Can computing costs be calculated for a single query/update statement? Also, is there a procedure for controlling usage from a single IP (malicious or otherwise)? Thanks....

Storage costs are typically based on the amount of data stored in your database. You can monitor this using Azure's built-in metrics and then allocate costs based on each client's data usage. Compute costs in Azure SQL Database are primarily based on the Database Transaction Unit or vCore model you are using. These models determine the amount of CPU, memory, and I/O resources allocated to your database. You can use Azure Query Performance Insight tool to monitor and analyze the performance of individual queries. Use DMVs like sys.dm_exec_query_stats if you want to ge detailed information about query execution, including CPU time, logical reads, and elapsed time. Also, Azure provides metrics such as CPU percentage, Data IO percentage, and Log IO percentage. These can be used to estimate the compute usage of your queries. To control or restrict access from a specific IP address, you can use : Firewall Rules Virtual Network Rules Threat Detection

Allocating DB costs - Microsoft Q&A

Accepted answer

Amira Bedhiafi 28,766 Reputation points

2025-02-18T21:52:59.79+00:00
Storage costs are typically based on the amount of data stored in your database. You can monitor this using Azure's built-in metrics and then allocate costs based on each client's data usage.

Compute costs in Azure SQL Database are primarily based on the Database Transaction Unit or vCore model you are using. These models determine the amount of CPU, memory, and I/O resources allocated to your database.

You can use Azure Query Performance Insight tool to monitor and analyze the performance of individual queries.

Use DMVs like sys.dm_exec_query_stats if you want to ge detailed information about query execution, including CPU time, logical reads, and elapsed time.

Also, Azure provides metrics such as CPU percentage, Data IO percentage, and Log IO percentage. These can be used to estimate the compute usage of your queries.

To control or restrict access from a specific IP address, you can use :

Firewall Rules

Virtual Network Rules

Threat Detection
Please sign in to rate this answer.
Admin eSMMS.tech 20 Reputation points

2025-02-19T13:36:52.97+00:00

Amira,

Many thanks for your quick and insightful response. I am new to this system and hopefully this comment will reach you.

If you are willing, I have some follow-up questions...

I am building a desktop client/server document management application that will use the DB to store shared content. I will need only about ten or so tables, and my typical user will have a couple of GBs of storage.

I initially tried Google drive, the attractive part being that users manage their own data. However, after much time and teeth-grinding implementing the Google APIs, code that was working simply broke. I could find no answers and no help.

Goodbye Google, hello Microsoft.

Btw, I currently develop in Visual Studio and have been using Microsoft since the early 80's. The integration with the Azure DB is terrific. I initially didn't use SQL because I thought I wouldn't be able to calculate per user costs. To my delight it appears (at least with storage) the costs can be easily calculated, and most importantly, are reasonable.

I set up my DB as Free General Purpose, Serverless, Gen5, 2 vCores.

I have spent some time looking at Intelligent Performance stats, but the compute costs have me confused. My initial development/testing exceeded monthly compute limits in a couple of days and I am not sure exactly why.

Is it because I left Visual Studio running with the DB connected?

Are there system processes running in the background that use CPU?

Just as with storage, I want to be able to predict and price per month compute costs for a user. Hopefully they are as reasonable as the storage costs.

I am willing to dig into the documentation, but there is SO MUCH and I want to spend the majority of my time in my code. If you can point me where I need to go it would be very much appreciated.

Admin eSMMS.tech 20 Reputation points

2025-02-19T17:12:15.7666667+00:00

Amira,

Thanks for your quick response. If you will indulge me, I have a couple of follow-up questions.

I am developing a desktop document management application that will use a SQL DB for content storage and sharing. I only need a few tables, and my typical client will use a couple of GBs for storage. My test DB is currently set up as "Free - General Purpose - Serverless: Gen5, 2 vCores".

I initially implemented my solution using Google. The attraction was that users would store their content on their own Google Drive with my App accessing it using the Google APIs. I spent a lot of time and teeth-grinding understanding and implementing the Google APIs. As I was about to deploy, one of the APIs stopped working. I could not find the reason or figure out how to get help.

Goodbye Google, hello Microsoft.

Btw, I have been developing code using Microsoft products since the early 80's. My app is built with C# using Visual Studio. I am impressed how easily it connects with Azure.

The reason I didn't use an Azure DB in the beginning was because I was concerned I wouldn't be able to determine per user monthly costs. After some investigation I found that per user storage costs are easy to calculate. I am still concerned with computing costs.

Per the suggestions in your answer above, I dug into Intelligent Performance for my DB and saw that query level stats are available. However, my DB was halted after a couple of days due to exceeding the compute seconds allocated for the month and I don't know why.

Is it because I left my DB connected and Visual Studio running?

Are there system processes running in the background using compute resources?

I don't want to be blindsided by computing costs, and want to communicate the potential costs to my users. Storage is easy - the user pays costs up front for the number of GB he wants. I can predict how much storage is needed and calculate the cost.

How could I predict, for instance, 6 months of computing costs?

I am willing to dig into documentation but there is SO MUCH it's a little overwhelming. After investing all that time with Google only to kick them to the curb, I'd rather find out sooner than later whether the MS model works for me.

I would very much appreciate it if you could help by either explanation or pointing to relevant documentation.

Thanks in advance...

Jeff
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Share via

Allocating DB costs

1 additional answer

Your answer