April 23, 2011 at 9:30 am
Hello,
One of my recent requirement is to provide my client with an 100% high availability solution for SQL Database.
Thinking about proposing a Clustering solution, 2 Cluster pairs with 2 independent SQL instances running on them, for eg
Cluster Pair 1:
Node A : 025 N1
Node B : 025 N2
SQL instance : 025SQL
Cluster Pair 2:
Node A : 026 N1
Node B : 026 N2
SQL Instance : 026 SQL
Principal database name : 'Remote db' present in 025SQL
Mirrored database name : 'Remote db' present in 026 SQL
How should i setup my witness in this scenario for favoring an automatic failover ? Can my Witness be just a standalone VM or Physical server, is there any additional disk\process requirement to set up a Witness server.
Do any of you have any alternate moves for setting up an 100% availability scenario, if so please do help.;HArdware concepts like EMC's SRDF etc or SAN based mirror will be taken from Hardware engineer perspective, am to suggest from DB end only.
I know Log ship is out of the box for 100% availability, if there is any difference to Remote based mirror on Cluster , help me know
Thanks
April 24, 2011 at 11:07 pm
It will be same as you create witness server in a normal server environment,in cluster environment you always use cluster SQL name,now scenerio is
Cluster 1 Name Primary : MSSQL_PRI
here is 2 Nodes Active/Passive
Cluster 2 Name Secondary : MSSQL_SEC
here is 2 Nodes also Active/Passive
you can create witness server on the PRIMARY cluster server as well as SECONDARY Node or create on a separate pc or server as you want,there is no high requirement for witness server because its use for heartbeat of a primary server nothing else
and if you developed an application which is dotnet 3.5 or later then you can define failover connection in the connection string of the application,this will work autmatically
Regards,
Syed Jahanzaib Bin Hassan
BSCS | MCTS | MCITP | OCA | OCP | OCE | SCJP | IBMCDBA
My Blog
www.aureus-salah.com
April 28, 2011 at 7:47 am
Thanks Syed, the info is useful
April 30, 2011 at 3:04 am
Syed,
If we have developed an application which relies on dotnet 3.5 or later then we can define failover connection in the connection string of the application, this will work automatically
Can you help me know on where and how to do the changes to application, where would this connection string be present, will the application adminsitrator be aware of doing this, does this have to do anything with ODBC etc.
Thanks
Eben
May 1, 2011 at 1:41 am
Data Source=myServerAddress;Failover Partner=myMirrorServerAddress;Initial Catalog=myDataBase;Integrated Security=True;
http://msdn.microsoft.com/en-us/library/ms366348.aspx
http://msdn.microsoft.com/en-us/library/ms130822.aspx
Regards,
Syed Jahanzaib Bin Hassan
BSCS | MCTS | MCITP | OCA | OCP | OCE | SCJP | IBMCDBA
My Blog
www.aureus-salah.com
May 4, 2011 at 3:02 pm
I am currently setting up a clustered mirrored server on a offsite collocation for DR.
The problem with having automatic failover for mirroring is that if the principal clustered server fails over to another node in the cluster the witness server will see this and then fail over to the mirror server. You might not want this behavior and would prefer the mirroring to remain on the principal server so we are not going to use the automatic failover. Here is an article you can read about your exact situation.
http://technet.microsoft.com/en-us/library/ms191309.aspx
-------------------------------------------------
Will C,
MCITP 2008 Database Admin, Developer
May 5, 2011 at 2:52 am
Yes thats a well made statement, offcourse when sysadmin is going to apply windows patches it can affect the database with bounce..
Eben
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply