Let's take these in opposite order, because I think there may be some
confusion.
First, a multi-instance cluster ( the correct name for "active-active") does
not offer any native scale-out capabilities. You have two independent
instances reading and writing to two sets of databases. If the second
instance can be "read-only" from the application's perspective, then you can
get some benefit.
As for the upgrade, there are certainly some compatibility issues that have
to be tested, identified, and remediated. Much of the time and effort on my
migration engagements focuses on compatibility. The actual upgrade is much
easier with SQL 2008 now that Microsoft added rolling upgrades as a feature.
Short version is you upgrade a passive node for each instance, failover the
instance to that node and then upgrade the remaining node(s). You always
have an available node to run SQL on if something goes wrong.
--
Geoff N. Hiten
Principal SQL Infrastructure Consultant
Microsoft SQL Server MVP
"TechNeedy" <TechNeedy (AT) discussions (DOT) microsoft.com> wrote
Quote:
I've assigned to suggest proposing a solution in which the stakeholders
are
interested in having their SQL upgrade from 2000 to 2008. They would also
like to have their clusters (2 different clusters) be Active / Active with
the 2nd one have transactional replication from the 1st cluster.
The expectation of having 2 clusters with Active/Active is to enable HA
when
the traffic is really spikes.
Pls Adv:
Potential issues & challenges in migrating from 2000to2008?
Clustering :- Is this a good design? what are the challenges?
how else this can be tackled?
thanks & appreciate your advice/suggestion. |