June 26, 2014 at 1:16 pm
Ok I am considering moving what is currently a win 2008 sql 2008r2 activve/passive cluster that does log shipping to two servers, one for DR, the other for day old read only reporting to -- SQL 2012 active/passive cluster and always on to two servers, one for DR the other for real time reporting. The reportin server has other db's on it so it will be replaced. Sounds reasonable. My ? is the way to get there. I have two approaches.
Upgrade win2008 sql 2008r2 cluster to sql 2012 active/passive.
Rebuild dr server to win 2012 and sql 2012, build new win 2012/sql2012 for reporting server
Add them as AG group to this cluster and I am the hero.
Limitations are, has to be done over weekend. (closed after noon saturday till 6 am Monday) No extra hardware other than new reporting server, have to live with what I have. I have heard issues about running sql 2012 and AG on a windows 2008 server.
2nd Approach
Take passive node out of win 2008 cluster on Friday (pray nothing breaks)
Build that node as win 2012 cluster
Rebuild dr server to win 2012 and sql 2012, build new win 2012/sql2012 for reporting server
Detach all db's on active node (on san storage) rebuild that node to win 2012 and add to 2012 cluster and add on sql 2012
Complete AG group setup db's on read-only and DR server and I am hero again.
Limitations has to be done over weekend. (closed after noon saturday till 6 am Monday) No extra hardware other than new reporting server, have to live with what I have. Failure is not an option. Whole company runs on this
Anyone tried this yet? Any input on either approach
June 27, 2014 at 8:55 am
Go for 2 solution. Just don't forget to take backups before performing upgrade.. I would advise not to detach, but restore databases on SQL2012SP2. Restore would assist system tables upgrade to 2012. Of course the best solution would be to build new 2012 cluster and perform migration, rather than trying to destroy the cluster and try to perform each node upgrade with taking some risk, of course. The problem would be in this scenario: shared storage presentation. You would need to get a strategy how it is going to be presented. Hope it helps and good luck.
~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*Brevity is the soul of wit.
June 27, 2014 at 8:56 am
Might be helpful
~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*Brevity is the soul of wit.
June 27, 2014 at 10:21 am
kind of leaning that way, this is a point of new return migration with about 36 hours to get every piece done, as soon as we take the first node out of the cluster and rebuild any of the log shipping servers I have to go forward. Worst is there is no testbed for this.
June 27, 2014 at 10:23 am
probably should have mentioned this is all physical
June 28, 2014 at 6:49 am
Personally we tried to perform SQL and windows upgrade for cluster, as client requested (2005 to 2008R2 SQL, 2003 to 2008R2 Windows server). Unfortunately we ended with up with building new platform for them and full migration.Cluster included DR solution. As you mentioned before, you already have log shipping established? SO what stops you to bring secondary on line ad rebuild the cluster from the scratched. It is more likely that something could go wrong.
~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*Brevity is the soul of wit.
June 29, 2014 at 1:54 pm
tcronin 95651 (6/26/2014)
Ok I am considering moving what is currently a win 2008 sql 2008r2 activve/passive cluster that does log shipping to two servers, one for DR, the other for day old read only reporting
Can you provide more detail please. How many nodes in the cluster, how many clustered instances?
What spec are the 2 log shipped secondarys, what spec are the cluster nodes?
What storage design do you have for the new Windows 2012 cluster, all local?
tcronin 95651 (6/26/2014)
to -- SQL 2012 active/passive cluster and always on to two servers, one for DR the other for real time reporting. The reportin server has other db's on it so it will be replaced.
Please detail more clearly your intended design for the 2012 system, how many nodes in the windows cluster, how many instances\replicas do you intend to have?
Will any of the sql 2012 instances\replicas be clustered instances.
tcronin 95651 (6/26/2014)
Limitations has to be done over weekend. (closed after noon saturday till 6 am Monday)
This is not a limitation, it's a luxury 😉
tcronin 95651 (6/26/2014)
No extra hardware other than new reporting server, have to live with what I have.
Again easily doable using existing hardware, it's even possible to leave the old system in place as a rollback.
tcronin 95651 (6/26/2014)
Failure is not an option.
It never is 😀
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply