November 1, 2007 at 11:38 am
I'm having a discussion with some architects and I want to make sure I haven't missed something, so any input you guys have is welcome. If you were replicating data from a production data center to a dr data center what issues would you have state as risks:
SAN REPLICATION: utterly depending upon the SAN Replication architecture
LOG SHIPPING: dependent upon network connection; no breaks to the LSN # of the log shipped tranlog; restrictions on backups so as to not break LSN#; no changes to recovery mode (like bulk loading)
REPLICATION: Dependent on data to be moved, frequency of updates and network connectivity
DATABASE MIRRORING: dependent upon network connection
Others?
November 1, 2007 at 3:15 pm
that pretty much sums it up.
it also depends on the dr data center...
You have to have the same exact type of hardware and software as in prod.
Disaster Recovery options depend on the up time and how much data are you willing to lose.
We use log shipping for all our sql servers and are behind one hour.
November 1, 2007 at 3:21 pm
Agreed that you summed it up fairly well, under the assumption that you are looking that the asynchronous versions of those. Anything synchronous would have a "risk" attached to it that the link to the remote location can in fact keep up, and doesn't go down during said disaster.
Even if asynch alleviates the issue - the link to the remote should probably always stay on the list as a "risk". No link = no DR coverage.
Technically, you could run a cluster over such a connection as well, but the reality is that the link tends to get saturated rather quickly.
----------------------------------------------------------------------------------
Your lack of planning does not constitute an emergency on my part...unless you're my manager...or a director and above...or a really loud-spoken end-user..All right - what was my emergency again?
November 1, 2007 at 11:37 pm
I see log shipping as the best choice for Dr if the customer accepts a little dta shortage. With
Replication , there is a chance that you may miss objects and data if they are not configured.
Dtabase mirroring, you will not be able to do a point in time recovery if an user accidently deletes a row
Clustering is constrainted geographically
My option is for LS.
Cheers,
Sugeshkumar Rajendran
SQL Server MVP
http://sugeshkr.blogspot.com
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply