• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 340
  • Last Modified:

Active/Passive Clustering not high availability

I currently have an active passive cluster to support my SQL 2000 OLTP environment. In a recent test of the cluster's fail over (I simulated a server failure by unplugging the nic on the active node) the system dropped. I have 3 applications that rely on the database to function, they all failed. The cluster service died and then restarted changing over the passive node.  I had to manually restart the applications to get back up and running. I spoke Dell (they setup the cluster) and they said that clustering wont provide what I am looking for. Can I get your thoughts on this? Will an active/active cluster provide this?
0
wildwoodslots
Asked:
wildwoodslots
2 Solutions
 
page1985Commented:
A failover cluster does not provide instantaneous failover.  When one node in the cluster failes, depending on your configuration, it takes from 15 seconds to a minute for the cluster to begin providing services again.

An active/active cluster would provide better failover times than an active/passive.  However, you may want to consider a load balanced cluster, as unlike a failover cluster, all nodes in a load balanced cluster respond concurrently at all times.  Meaning, if one node fails, all the others are already active and respinding anyway, so the failed node gets evicted until it comes back online.
0
 
markpalinuxCommented:

An Active/Active cluster for SQL 2000, means that databases and a sql instance is live on both nodes. However that is very difficult if your users read and write to the databases.  Also it increases the licensing fee from MS, as it doubles the number of Active nodes.


Your cluster should be setup like:
node 1 - node 1 ip address, node 1 network name
node 2 - node 2 ip address, node 2 network name
sql1 - SQL instance - with its own ip address., SQL instance network name

be sure your apps are using the SQL instance name and not the node1 or node2 names.

the sql1 sql instance name and ip address should become active on the "passive" node when a failover occurs.

In cluster admin you should be able to see which nodes are up, which node owns a group.

When I am running failover testing for a cluster I keep open a ping window with "ping -a clustername" so I can see just how many packets drop during switchover.

As far as the cluster service stopping review the event logs. sounds like this maybe a new environment for you maybe upgrade would be a good idea.

Mark

0

Featured Post

Concerto's Cloud Advisory Services

Want to avoid the missteps to gaining all the benefits of the cloud? Learn more about the different assessment options from our Cloud Advisory team.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now