SQL Server Performance Forum – Threads Archive

Active/Passive to Active/Active

We would like to move from Active/Passive to Active/Active Configuration. We have the following hardwares: 2 compaq proliant DL760
8 Processor on each server
8 gb on each server
SAN coneected with fibre channel My question is: a) We have around 1 Terrabytes of data. Do we need to keep two copies of data for each nodes ? If so, do we need to do merge replication to keep the data insync?. b) We have log shipping going to the standby server. How do we setup the logshipping if we have two copies of data ? Is there any documentation available for active/active config ? Thanks,
Gopal

An active/active install is just two active/passive installs, one after the other. For an active/active install, each instance of SQL Server will have to have its own dedicated shared array space. So if you only have a single shared array now, it cannot be used for active/active until you either split it, or get another array. SQL Server clustering is for fault tolerance, not load balancing, which is what it appears you want to do. SQL Server load balancing has to be done through third-party products. Trying to use replication to synch data between two databases won’t work the way you expect, and it is fraught with many problems. I guess if you give us more info on what you want to accomplish, we can try to offer some assistance. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

Thanks Brad for your email. We have Active/Passive Cluster and we want to change to Active/Active Cluster. We currently have 1 TBs of data. The reason for this change is, we want to have 100% uptime. In Active/Passive Cluster, we still have 99% uptime but the management feels we can go for Active/Active Clustering. I want to present the management with the benefits what we get out of Active/Active Cluster. I have the following questions. a) Since we have to maintain two copies of data in Active/Active, if one of the node is offline how the data will be insync with the both the nodes ?
b) How is the logshipping works on Active/Active ? If one of the node is offline, how it works ? I would appreciate, if you could help me on this. Thanks,
Gopal

quote:Originally posted by gopal
a) Since we have to maintain two copies of data in Active/Active, if one of the node is offline how the data will be insync with the both the nodes ?

There is no automatic way that data will be kept in sync. It’s not different from keeping data in sync between two single servers. You have to set up some form of replication but as brad mentioned there are risks with that and performance can take a hit as well, higher maintenance will also be required and it’s not sure that the database design you have will work with for example merge replication.

quote:Originally posted by gopal
b) How is the logshipping works on Active/Active ? If one of the node is offline, how it works ?

It’s not different from loggshipping between two single servers where you shut down one node. You have to handle the event if the target server is down. Also see this post for more info on "active/active":
http://www.sql-server-performance.com/forum/topic.asp?TOPIC_ID=949 /Argyle

Going to active/active won’t help your reliability, as it will not affect the reliability of the shared array. The only advantage of active/active is if you want to get more work out of your present hardware resources. In fact, I personally believe that active/active is less reliable than active/passive, as there is more to go wrong. What we do, is run active/passive, then log ship the data to a warm backup server at an off-site location. This way, if the cluster or database goes bad, we can be back up and running in 30 minutes or less, with minimal loss of data. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

]]>