October 29, 2010 at 3:04 am
Hi,
I would like to configure an HA environment 24 x 7 x 365 availabily with SQL Server 2008.
I have already informed about the options that SQL Server 2008 give to me.
The one that best suits my needs probably would be an ACTIVE/PASSIVE clustered configuration, but the only problem i can see is that in case of active node failure there is a delay from 10 to 20 seconds until the SQL Server is newly operative.
With witch SQL Server 2008 configuration can i configure a environment (without loss of data) in witch response time (in case of failure) was 0? is possible?
Thank you very much in advance.
October 29, 2010 at 3:12 am
If you need that kind of availability and you're not familiar with HA/DR, get a consultant in to advise and help you. This is not an area you want to be trying to get right without experience.
Clustering's can be a good place to start, but it's not sufficient, you'll need a secondary failover at least for the case where the entire primary site is unavailable.
Absolute 0 downtime is not possible with any failover solution. There's always going to be some small time to cut over to the secondary.
Typically the closer you want to get to 24x7x365 with the 5 nines uptime, the more you're going to be paying to get said solution, the more work it's going to take to get to it..
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
October 29, 2010 at 4:58 am
you re not going to reach 0 downtime therefore you got the 9's in you sla. Also take in account which application you use you re db server for
Consider:
Clustering, even with dual dc with san replication to be not site depended
mirroring with with witness.
or even clustering with mirroring if not dual dc can be implented
also polyserve can be an solution.
like gail said the more you pay the more is possible. but 0 response time you wont get either from the SQL server or the reconnection of you re applications to the sql databases.
October 29, 2010 at 11:51 am
database mirroring in High Safety mode is the only thing that comes close. clustering is a waste because there is still a single point of failure... (the SAN)
I have been involved in building many HA enviroments, and stretch clustering with mirrored SANS is considered the "Holy Grail", however, in order to administor a GEO cluster with mirrored SANS is extremely difficult and failover time is always more than a standard cluster with out SAN mirroring.
you will achieve the highest availibilty with Database Mirroring in high safety mode. typical fail over time, if your app supports it, can be under 3 - 5 seconds.
it can be simplified like this; (the times are averaged from my experience, everyones will be different)
failover time for mirroring = less than 10 seconds
failover time with clustering = less than 1 minute more than 30 seconds.
failover time with GEO clustering and SAN based mirroring = less than 3 minutes, more than 1 minute.
October 30, 2010 at 2:39 am
Geoff A (10/29/2010)
failover time with clustering = less than 1 minute more than 30 seconds.
I have deployed previous clusters that can failover 2 instances of SQL server in around 15-20 seconds,
A simple thing like an incorrect NIC binding can slow down things drastically!
Geo clusters are IMHO best for the ultimate DR protection but can be hugely expensive and complex, whereas mirroring is already built in.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply