Azure-docs: When will this come out of preview?

[clement911]

We are very keen to use this. When will this be released?

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: 7a2961d1-4448-11b7-2dc2-3b738bd70939
• Version Independent ID: 206eb022-25c4-66d7-7285-0d53a84956e3
• Content: Azure SQL Database - read queries on replicas
• Content Source: articles/sql-database/sql-database-read-scale-out.md
• Service: sql-database
• GitHub Login: @anosov1960
• Microsoft Alias: sashan

[Mike-Ubezzi-MSFT]

@clement911 Thank you very much for your interest in Azure products and services. We are actively investigating and will get back to you soon.

[CHEEKATLAPRADEEP-MSFT]

@clement911 Regarding service general availability, I would suggest to keep an eye on Azure updates.

Azure updates provide information about important Azure product updates, roadmap, and announcements.

[clement911]

I will keep an eye on it, but is there a rough date for this feature?
It's been in preview for ages! Many months.
We've hold back using it because it's not clear what the limitations are and we don't want to cause any potential downtime to our app.

[CHEEKATLAPRADEEP-MSFT]

@clement911 Currently, we don't have any ETA for the general availability.
We will now proceed to close this thread. If there are further questions regarding this matter, please comment and we will gladly continue the discussion.

[clement911]

Are there a list of known risks/limitations of the current preview?

[CHEEKATLAPRADEEP-MSFT]

@clement911 During preview, these are the following limitations:
• Query Data Store and Extended Events are not supported on the read-only replicas.
• Replication latencies within the region are low and this situation is rare.
• At any given time only one of the AlwaysON replicas is accessible by the ReadOnly sessions.
• Round-robin or any other load balanced routing between the local replicas of the secondary database is not supported.

[clement911]

Thank you for that.

Replication latencies within the region are low and this situation is rare

What situation?

[CHEEKATLAPRADEEP-MSFT]

One of the benefits of replicas is that the replicas are always in the transactionally consistent state, but at different points in time there may be some small latency between different replicas. Read Scale-Out supports session-level consistency. It means, if the read-only session reconnects after a connection error caused by replica unavailability, it can be redirected to a replica that is not 100% up-to-date with the read-write replica. Likewise, if an application writes data using a read-write session and immediately reads it using a read-only session, it is possible that the latest updates are not immediately visible. This is because the transaction log redo to the replicas is asynchronous.

[clement911]

Got it, that sounds totally reasonable.

[clement911]

At any given time only one of the AlwaysON replicas is accessible by the ReadOnly sessions.

Will this be fixed for the final release? i.e. round robin between the different replicas?

Overall it sounds like preview would be pretty stable and this limitations seem not too bad.

[jstuparitz]

Any update on this becoming GA? Can you go GA without the round-robin? We'd very much like this use this capability in a GA setting.