How can we improve Azure Storage?

Ability to manually initiate failover on Geo-Redundant storage accounts (GRS & RA-GRS)

GRS & RA-GRS storage accounts currently geo-replicate all contents from a primary zone's storage account (storageA) to a secondary zone's storage account (storageB) automatically and this is wonderful.

Right now, it is Microsoft that control when the failover occurs from storageA to storageB. It is also very well documented that this switch will happen only as a last resort move because of the complexity of switching all accounts, possible data loss, etc...

For advanced DR scenarios it would be very handy for us to have a way to manually trigger the failover switch between storageA and storageB.

After the manual switch, DNS would point to storageB instead of storageA and read/write operations would work as expected on the newly promoted storage account (storageB)

226 votes
Vote
Sign in
(thinking…)
Sign in with: oidc
Signed in as (Sign out)
You have left! (?) (thinking…)
Stephane Lapointe shared this idea  ·   ·  Flag idea as inappropriate…  ·  Admin →

15 comments

Sign in
(thinking…)
Sign in with: oidc
Signed in as (Sign out)
Submitting...
  • Javier Bello commented  ·   ·  Flag as inappropriate

    Hello team do you know if is a estimated of time where can this available for Central US & East US 2 regions? Thanks

  • Ramsey Callaway commented  ·   ·  Flag as inappropriate

    After yesterday's South Central US outage, this ability is more important than ever. Microsoft is unwilling to call a failover event at the risk of having to update all of their marketing slides, so we need this power in our hands.

  • Jason Tang commented  ·   ·  Flag as inappropriate

    Hi guys, we are actively working on customer controlled failover scenario but don't have a release date to share it. If you are interested in this scenario, please don't hesitate to contact me at Jutang@microsoft.com so we can learn more from you, thanks.

    -Jason Tang (PM of Azure Storage)

  • Shaun Laughton commented  ·   ·  Flag as inappropriate

    Tis really is a glaring omission. I may be overlooking something, but this kind of feature is available for Azure SQL Database in the form of Failover Groups. Why not storage?

  • Gonzalo Parra commented  ·   ·  Flag as inappropriate

    This is an important requirement so we can have control over the RTO, otherwise its a big question mark...

  • Krishnendu Ghosh commented  ·   ·  Flag as inappropriate

    If a certain amount of data loss is acceptable in a particular scenario and there is short RTO requirement, manually initiating a failover is highly required feature.

  • Gordon commented  ·   ·  Flag as inappropriate

    It must be supported, it has no point that failover is out of control on customer side.
    Failover is not only for the purpose of "Real Disaster" happens.

  • Ranger Due commented  ·   ·  Flag as inappropriate

    CSA: Without the ability to test this feature, it greatly puts into question the ability of my clients to correctly measure their risk profile on Azure. It is an unnecessary blocker.

Feedback and Knowledge Base