I've been reading my Sql Server 2005 Maintenance and Implementation book on
Mirroring and Michael Hotek gives an example of a client that required
99.995% availability as well as zero data loss except in the event of a
natural disaster. They also required automatic failover and transparency to
the apps.

His comments were that "clustering definitely could not meet the downtime
requirements, although it could meet nearly all the other requirements if we
made a couple of simploe changes to the data access object to add reconnect
logic".

I'm not that familiar with clustering, so I was wondering why it could not
meet the above stringent downtime requirements? Also, any comments you have
about the comparison between clustering and mirroring, I would appreciate.

I am going to make an assumption here that the 99.995% is for all time and
does not allow for scheduled maintenance windows. As such, you can apply
maybe one SQL service pack every two years, given no other failures. With
Mirroring, you move everything to an alternate host during maintenance
windows with only a few seconds of inactivity and no need for a client
reconnect. Also, clustering does not protect against disk corruption or
failure so you have to factor that in.


--
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP





"CLM" <CLM (AT) discussions (DOT) microsoft.com> wrote