Data redundancy for Storage and Database


  • Hello,

    I have 2 questions:

    1- If I use RA-GRS redundancy for Storage, is there a need to backup the storage?

    2- In Azure SQL Database, are there additional costs for active Geo-replication, is it integrated in this service?

    Thank you! 

    Friday, June 10, 2016 11:22 AM

All replies

  • Hello,

    Thank you for posting your questions on Microsoft TechNet.

    For Q1. Microsoft Azure Storage Redundancy is fine for availability, but you need a good backup incase of data corruption or accidental deletion of data.
    Here you find more information about Microsoft Azure Backup :

    Technical Documentation :

    Backup Pricing :

    For Q2. Here you find Microsoft Azure SQL pricing :

    Hope this information is useful for you.

    Best Regards,


    (Ps. If this information is helpful for you, please mark your questions as answered for the community.)

    Friday, June 10, 2016 12:06 PM
  • Hi sna_3444,

    1. It depends on what you're storing. Just making sure you have a secondary copy of your data is always a good thing. A couple of things to know: There is a lag with no SLA between the time the data is written to primary storage and the time it is replicated to secondary storage. Also, the order in which the data is replicated isn't guaranteed. This is a great article explaining some of the intricacies involved in the geo-replication:  My point is that reading the secondary copy, you may miss out on some data that hasn't been replicated yet.

    Second thing to know: if you are running VMs (i.e. you are storing VHD files and replicating them), be aware that you can't just copy that .vhd file from the secondary storage account to another location and create a VM from it. There's a slim chance it might work, but there's certainly no guarantee. There are two issues. For one thing, the VM writes to storage continuously, so there's no telling what state the VM will be in when you try to create it from the VHD file. You need what's called a "consistent" image, and to get that, you have to take a snapshot while nothing is writing to any of the attached disks (including the OS).

    To get a consistent snapshot, you either have to run VSS (Volume Shadow Copy Service) for windows or shut down the VM and create a snapshot of each VHD file. You can then bring the VM back up and let it run, and copy the snapshots to blobs and create the VM from them.

    2. When you pay for Azure SQL Database, all costs are included. If they offer geo-replication, then the price should be included in the service. This is different from running SQL Server in a VM.


    Sr. Content Developer at Microsoft

    Thursday, June 30, 2016 11:40 PM