May 28, 2011 at 10:08 am
Syed Jahanzaib Bin hassan (5/28/2011)
Active/Active means in term of microsoft multiple instances on mutiple nodes and it follow round robin algorithm for fail over,there is no any Load Balancing like Oracle RACActive/Passive means,fail over only if 1 node down then services start on another node automatically
If you have required Logshipping and database mirroring on another node then you have required
Active/Active cluster of microsoft SQL server,
1 instance for Primary Database
1 instance for logshipping or database mirroring
That is only an option but not necessarily required. Mirroring or logshipping to a different server in the same cluster will work but hardly provides real protection. Be careful of configuration choices such as this.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
May 29, 2011 at 2:01 pm
I'm with SQLRNNR on this, log-shipping or mirroring to other machines in the same cluster doesn't do much for either high-availability or disaster-recovery. With that said there are other reasons to use those technologies that makes sense, such as a reporting server. What is important is understanding your goals in an HA solution and understanding if the solution proposed will satisfy those goals.
If load balancing is a goal then clustering is probably not the solution.
CEWII
May 29, 2011 at 11:21 pm
yes offcourse,I didnt mentioned that is the only solution
Regards,
Syed Jahanzaib Bin Hassan
BSCS | MCTS | MCITP | OCA | OCP | OCE | SCJP | IBMCDBA
My Blog
www.aureus-salah.com
May 31, 2011 at 6:21 am
clustering isn't always an advantage - look at the cons for clustering. A while ago at a technet meeting at Microsoft we did a straw poll on whether people thought clustering saved or created issues, the majority thought that clustering gave more problems than it resolved. Most failovers were considered to be a "feature" of clustering and if the server had been standalone would not have happened.
e.g. a network connection drop may well cause a cluster failover which may or may not cause chaos to your application(s) .. on a standalone server a dropped network connection will not cause a server reboot!
There is sometimes a blind faith that you must have clustering, in 5 years my clinet has never had a node failure on any of their clusters ( e.g. server hardware fault ) but we've had lots of failovers - usually caused by networking ( ethernet and fibre channel ) "issues" -- notwithstanding the odd failover which no-one can explain < grin >
[font="Comic Sans MS"]The GrumpyOldDBA[/font]
www.grumpyolddba.co.uk
http://sqlblogcasts.com/blogs/grumpyolddba/
May 31, 2011 at 6:26 am
WayneS
So your saying if I did a 2 node failover and did log shipping that would be considered a 3 node cluster by microsoft?
May 31, 2011 at 8:27 am
Thanks for all of your replies! I have my answer: Active/Active would be necessary for my clustering needs.
May 31, 2011 at 10:20 am
bopeavy (5/31/2011)
WayneSSo your saying if I did a 2 node failover and did log shipping that would be considered a 3 node cluster by microsoft?
No but it would be high availability and Disaster recovery - so it could be a good solution.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
May 31, 2011 at 10:20 am
bopeavy (5/31/2011)
Thanks for all of your replies! I have my answer: Active/Active would be necessary for my clustering needs.
Awesome - you're welcome.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
May 31, 2011 at 10:55 am
WayneS or SQLRNNR
Not sure of that, But I read the licensing agreement from Microsoft and they consider it to be part of a failover cluster in my case of active/active + Log shipping would be consider a 3 node cluster. What are your thoughts? I have included the link from microsoft below under PASSIVE SERVERS / FAILOVER SUPPORT.
May 31, 2011 at 11:19 am
bopeavy (5/31/2011)
WayneS or SQLRNNRNot sure of that, But I read the licensing agreement from Microsoft and they consider it to be part of a failover cluster in my case of active/active + Log shipping would be consider a 3 node cluster. What are your thoughts? I have included the link from microsoft below under PASSIVE SERVERS / FAILOVER SUPPORT.
I doubt it would be considered a 3 node cluster, log shipping and clustering have really nothing to do with each other, can you point out the section? I have read that doc a number of times and never got that out of it. You would need a third license but it wouldn't need to enterprise edition. SQL Standard supports log shipping in 2008, per: http://msdn.microsoft.com/en-us/library/cc645993.aspx#High_availability.
CEWII
May 31, 2011 at 11:23 am
Thanks, Elliott!
May 31, 2011 at 11:34 am
You are very welcome.
CEWII
June 1, 2011 at 7:25 am
bopeavy (5/31/2011)
Thanks for all of your replies! I have my answer: Active/Active would be necessary for my clustering needs.
the cluster A/A is the simple concept!!! you will have instances on each node where the resources like: memory, cpu don't will shared!!! because this is concept of Windows of High Availability or HA, that is different of Oracle RAC or GRID!!!
Now, if your problem is performance then you have to use Windows HPC Server, it's a concept of High Performance!!!
link: http://www.microsoft.com/hpc/en/us/default.aspx
June 1, 2011 at 7:56 am
rfr.ferrari
No problem on performance this is for High availiability being able to recover with low or no loss of work time. In the case of hardware failure active/active failover in the case of a storage array failure log shipping to another server that would be up in a 1/2 hour after job scripts run against that server to bring everything online. Just part of a D-N-R strategy.
June 1, 2011 at 8:51 am
ok, alright!
but you have should in mind that each node only can work with half workload, because in point of failure, the active node will receive all workload divided in 2 servers!!!
Viewing 15 posts - 16 through 29 (of 29 total)
You must be logged in to reply to this topic. Login to reply