AlwaysOn performance on Replica databases

Question

AlwaysOn performance on Replica databases

Thomas Mucha

SSC Eights!

Points: 885
More actions
September 26, 2013 at 9:08 am

#395562

I am looking into AlwaysOn Availabilty Groups as both an HA and load spreading solution. But there is something I would like to clarify before I present to management.
Our production databases suffer from lots of blocking and deadlocks. Will this resource contention be "replicated" on the replica databases when the logs are applied? I'd like to offload many of the read operations (adhoc reports, etc.) to the replicas but don't just want to end up in the same situation as I am in now.
Thanks for any help.
Tom
p.s. - work is underway to address deadlocking, but it won't be fixed any time soon 🙁
In SQL there are no absolutes, it always depends...

Viewing 5 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply

Gail Shaw SSC Guru Points: 1004504 More actions · Answer 1

http://www.sqldiablo.com/2012/08/01/alwayson-availability-groups-isolation-levels-selects-blocking-writers/

Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability

We walk in the dark places no others will enter
We stand on the bridge and no one may pass

TheSQLGuru SSC Guru Points: 134017 More actions · Answer 2

Thomas Mucha (9/26/2013)
I am looking into AlwaysOn Availabilty Groups as both an HA and load spreading solution. But there is something I would like to clarify before I present to management.
Our production databases suffer from lots of blocking and deadlocks. Will this resource contention be "replicated" on the replica databases when the logs are applied? I'd like to offload many of the read operations (adhoc reports, etc.) to the replicas but don't just want to end up in the same situation as I am in now.
Thanks for any help.
Tom
p.s. - work is underway to address deadlocking, but it won't be fixed any time soon 🙁

It has been my experience consulting with numerous clients that deadlocks and blocking are most often rather easily addressed by proper indexing and it usually isn't very hard to figure out what index(es) to put where to accomplish the objectives.

Personally, given the hardware and licensing costs (and quite possibly consulting costs if you get help to get AGs done up right) to do SQL 2012 EE AGs, I would REALLY recommend doing a tuning exercise FIRST. It is quite possible that you can kill FOUR birds with one stone: faster queries, less blocking AND fewer deadlocks AND saving money because you won't NEED new hardware/EE licenses.

Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service

Thomas Mucha SSC Eights! Points: 885 More actions · Answer 3

TheSQLGuru (9/26/2013)
It has been my experience consulting with numerous clients that deadlocks and blocking are most often rather easily addressed by proper indexing and it usually isn't very hard to figure out what index(es) to put where to accomplish the objectives.
Personally, given the hardware and licensing costs (and quite possibly consulting costs if you get help to get AGs done up right) to do SQL 2012 EE AGs, I would REALLY recommend doing a tuning exercise FIRST. It is quite possible that you can kill FOUR birds with one stone: faster queries, less blocking AND fewer deadlocks AND saving money because you won't NEED new hardware/EE licenses.

Thanks for the reply. I didn't fully explain the scenario, sorry :ermm:

Hardware and licenses aren't really an issue as we will be using existing SQL instances that are currently logshipping secondaries. And, as mentioned, we are addressing the performance issues at source. This is legacy code, rife with nested cursors and repetitive/overlapping queries, that is being slowly reworked. Index tuning and maintenance is one of my main tasks which I carry out on a weekly basis.

The implementation of availability groups is being planned as part of a wider improvement to high availability and disaster recovery. And, I thought it would be advantageous to further take advantage of the ability to perform read-only tasks (backups, some reporting) from the secondary servers. it is just likely that AG will be implemented before the code rewrite. So I didn't want to make outlandish promises if the secondary databases will run into the same blocking/deadlocking problems as the primary ones.

I hope this makes sense.

Tom

In SQL there are no absolutes, it always depends...

Thomas Mucha SSC Eights! Points: 885 More actions · Answer 4

Thanks for the link.

In SQL there are no absolutes, it always depends...