November 19, 2020 at 7:38 pm
I have configured failover clustering and Always on AG group and planning to test the failover using both. So the plan is to test which would be the faster in terms of failover? Do you agree this would be the best approach or any other testing would need to take place? Please advise?
November 20, 2020 at 11:53 am
Sorry, I am not sure of your questions:
So the plan is to test which would be the faster in terms of failover?
Are you asking us if this is a good test or telling us that you plan to test if the AG or FCI failover is the faster?
Also:
Do you agree this would be the best approach or any other testing would need to take place? Please advise?
This question is quite vague so it is difficult to answer. The only advice I can think of is to make sure that you test using your applications. The fact that the AG or FCI failover completes doesn't guarantee everything works correctly.
November 20, 2020 at 1:02 pm
So you have two setups,
one the traditional failover cluster, shared storage
the other an AOAG, separate storage
The AOAG "should" be quicker to failover as all that is happening is moving the listener from A to B.
FCI's take longer as the service had to stop, storage has to be moved and the services restarted on the other machine.
But your question leaves a lot to the imagination as there is no detail at all about what it is your actually trying to do
November 23, 2020 at 6:19 pm
Thanks for the info. That helps.
Regarding storage. Is there any difference going with SAN or Local disk for AOAG?
November 24, 2020 at 8:00 am
If your using physical tin servers then you would want everything as close to the server as possible, local disk would be better if its the right type of local disk, you need disk which is going to be faster than the backbone link to the SAN and the SAN IO that can be provided.
This also then makes increasing disk space harder with local disk as you have to procure disk and that can then lead to having to evict replicas to add in storage, re-sync, rinse and repeat so there is more maintenance overhead when you have reached the disk capacity for local disk.
If your on VM's then is SAN storage anyway, just watch out for doing RDM disk as they can be hard to manage and limit things like snapshotting the VM etc.
These days though there is really no difference between local vs SAN, especially when you see things like the magic that Pure Storage are doing. Choose the right SAN at the right time and your laughing
December 1, 2020 at 12:40 am
Thanks. I was trying to add Polybase feature on existing traditional failover cluster and I get below error message
"Install of feature SQL_Polybase_Core_Inst is not supported on clustered instance".
Does polybase is not supported on traditional failover cluster? Please advise?
December 1, 2020 at 7:28 am
That is a new issue which should be in its own topic.
But as you have said yourself, you cannot install Polybase into a FCI.
To have multiple Polybase instances you must use the standalone installation and use the scale out feature
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply