sql-serverazureazure-sql-databasedatabase-restorepitr

Azure SQL database point in time restore taking over 24 hrs


I have been trying to do a PITR of a 2GB S0 Azure SQL Server db. It has been running for over 24hrs. The DB restore progress has been saying 50% complete for 18 Hrs without any errors. Should I upgrade the server DTUs and size or the actual service tier?


Solution

  • According to this post. On SQL Database, the "horsepower" is measured by Database Throughput Units, or just "DTUs". This unit is measured by an integer and may variate from 5 to 1750. Every database edition has an offer of one or more "Service Objectives", which are directly related to the number of DTUs and the price to be played.
    In the following image, you can find the list of "Service Objectives" (S0, P3, Basic, P11, S3, etc…) per SQL Database Edition and its respective prices. Notice that Microsoft is always updating its offer, so those prices and Service Objectives per Edition may be outdated when you read this post: enter image description here

    One option is a more conservative, responsible and dignified way to choose the number of DTUs, and is based on real data about your database activity. It is the DTU Calculator (http://dtucalculator.azurewebsites.net/), an online service that helps us by advising about the most appropriate Service Objective for a database. You just need to download a PowerShell script, available on the DTU Calculator website, and run it in the server where your database is located. As soon as you run this script, the following data will be measured and recorded in a CSV file: