asked on

How do I cluster 2 servers without using a single SAN or shared drive

Good day, Fairly new to Server 2008 and working on my SQL 2008 studies, however, we are a bit stuck. We currently have 2 Server 2008 R2 Enterprise servers at our datacenter, currently in DEV not live. What we need to do is have each server be HA without using a single point of failure with a single disk (SAN). Each server must have its own SAN in case of a disk failure. Additionally, we need to have our clients point to a single IP/DNS, which I assume would be a Virtual IP in either a cluster or NLB.
However, in all my research it seems that the only way to cluster is with a shared SAN. Can anyone help with avoiding a single point of failure with HA?

Thank you!
Paul

Paul Solovyovsky

You can't do SQL clustering without a SAN with 2008 but you can do this with SQL 2012 in the same way Exchange does DAG

jmichaelpalermo4

ASKER

Okay thank you, what about on the windows side OS clustering? I see you mentioned SQL 2012, currently we plan on a mirroring at the Database level Synchronized, with HA. Its the Server side that we will need the failover option.
Can you direct me to a link that will guide me through the process?
Will server 2012 OS allow DAG as well?
and will that accommodate our needs?
Maybe the best question is, can we cluster just the SQL and not Windows server?
If so, how does the client know when its pointing to to i.e.Server1\Instance1, upon failover the database will failover to Server2\Instance1, know to point to Server2?

Thank you!
Paul

Paul Solovyovsky

With Windows 2012 there are at least two nodes in the cluster so there is on need to cluster OS since each node will take over for the other one.

The link below has the difference between the two.

http://msdn.microsoft.com/en-us/library/ff929171.aspx

jmichaelpalermo4

ASKER

Thank you, I have read through this. Let me clarify my question. I need my servers to be redundant and must use WSFC in order to use HA clustering for SQL. I get that. My issue is, how can I create WSFC on 2 servers with 2 different SAN's. I need to know that if one server fails or the SAN fails, it will fail over to another server to allow for HA.

That being said, lets walk through this. Currently running server 2008 R2. Server 1 has SQL installed and instance1. Server 2 has SQL installed and instance1 as well. Currently my thought is mirroring the DB's, so if server 1 fails, server 2 becomes the principle... etc. Now how does the "Client" that is pointing to Server 1, switch to using Server 2. What tells the client that Server 1 is dead and needs to look at Server 2? Is there another way to create true HA, if both the SAN and server, or even if the SAN fails and WSFC that uses the same SAN to run the servers fails were down for both server.....

Which I may be completely off here thinking that WSFC clusters the disks where the OS lives. Maybe in my reading for the past 2 weeks I am missing some key pieces.

Here is what I am envisioning.

Paul Solovyovsky

If you're using Microsoft Clustering on 2008 than it will fail since databases are on the same SAN and SAN replication is usually async unless there are SAN technologies to compensate. If you're using SQL 2012 Distributed Groups than it doesn't matter you can have on second SAN or local storage since it's synchronous replication

jmichaelpalermo4

ASKER

Thank you for the reply. Are you referring to Availablity Groups? I did not find Distributed Group in the documentation

Paul Solovyovsky

Yes..keep thinking of DAG in Exchange..

jmichaelpalermo4

ASKER

Okay, based on my research, Always On requires WSFC to be installed first, is that correct?

Paul

ASKER CERTIFIED SOLUTION

Paul Solovyovsky

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial