99.99% without a cluster ?

Is it realistic expect 99.99% reliability and system uptime without using
any clustering solutions on a SQL 2000 db ?
My server hardware is dual proc, raid1=OS, raid5=db, dual nic teamed and
connected to individual switches for redudancy. My organization is 8 h/day
shop and I have plenty of time for maintenance during weekends or nights.
I am asking because some co-workers would like to cluster 4 SQL servers and
I have some questions if that is worthwhile and necessary in my environment.
Clustering is for hardware failures. If your motherboard or processor dies
you are down until you can replace it. If you can do that within a time
that is acceptable to your business then great, otherwise something like
clustereing or log shipping will get you going a lot faster.
Andrew J. Kelly SQL MVP
"Marlon Brown" <marlon_brownj@.hotmail.com> wrote in message
news:OF4IjVMLEHA.2576@.TK2MSFTNGP12.phx.gbl...
> Is it realistic expect 99.99% reliability and system uptime without using
> any clustering solutions on a SQL 2000 db ?
> My server hardware is dual proc, raid1=OS, raid5=db, dual nic teamed and
> connected to individual switches for redudancy. My organization is 8
h/day
> shop and I have plenty of time for maintenance during weekends or nights.
> I am asking because some co-workers would like to cluster 4 SQL servers
and
> I have some questions if that is worthwhile and necessary in my
environment.
>
|||It's "possible", but not "realistic" unless you're a zen master (:
You need to ask (at least) these questions:
(a) what is our REAL availability requirement
(b) what is our real cost of downtime
(c) what solutions are available to me (there is more than just clustering)
(d) what are the TOTAL costs associated with deploying each solutions?
(e) which solution (c) costs least (d) to limit downtime (b) whilst
achieving at least (a)
Regards,
Greg Linwood
SQL Server MVP
"Marlon Brown" <marlon_brownj@.hotmail.com> wrote in message
news:OF4IjVMLEHA.2576@.TK2MSFTNGP12.phx.gbl...
> Is it realistic expect 99.99% reliability and system uptime without using
> any clustering solutions on a SQL 2000 db ?
> My server hardware is dual proc, raid1=OS, raid5=db, dual nic teamed and
> connected to individual switches for redudancy. My organization is 8
h/day
> shop and I have plenty of time for maintenance during weekends or nights.
> I am asking because some co-workers would like to cluster 4 SQL servers
and
> I have some questions if that is worthwhile and necessary in my
environment.
>
|||What do you mean by 99.99%? Are you counting total uptime, or just
unplanned outages? We don't have 99.99% uptime measured as a total, even
with a cluster and I'm not worried at all. That's not to say that our
servers (Dell 6650's) aren't reliable, they are; we've only had one
unplanned outage in the past year and a half and that was only for about 2
minutes as the cluster failed over... However, between hardware
maintenance/upgrades, and patches--both OS and SQL Server, we have managed
about 99.93% total uptime. But we have managed 100% uptime for business
hours over the past 11 months.
What we have found to be the biggest benefit of clustering is the ability to
do rolling upgrades, thereby reducing the total outage window for any
particular upgrade.
"Marlon Brown" <marlon_brownj@.hotmail.com> wrote in message
news:OF4IjVMLEHA.2576@.TK2MSFTNGP12.phx.gbl...
> Is it realistic expect 99.99% reliability and system uptime without using
> any clustering solutions on a SQL 2000 db ?
> My server hardware is dual proc, raid1=OS, raid5=db, dual nic teamed and
> connected to individual switches for redudancy. My organization is 8
h/day
> shop and I have plenty of time for maintenance during weekends or nights.
> I am asking because some co-workers would like to cluster 4 SQL servers
and
> I have some questions if that is worthwhile and necessary in my
environment.
>