Active/Passive Clustering Scenario | SQL Server Performance Forums

SQL Server Performance Forum – Threads Archive

Active/Passive Clustering Scenario

Hello,
In an active/passive clustering scenario when the node1(Active) failsover to node2(secondary)do the sqlserver stop and start and i think the transactions running will
fail but does sqlserver gets restarted. Thanks,
You cannot start a failover cluster automatically using SQL Server. You must use Cluster Administrator in MSCS to automatically start a failover cluster. _________
Satya SKJ
Moderator
SQL-Server-Performance.Com
Not sure I understand the question but yes the cluster will try and start up the sql server service on the second node if it fails on the first node. It’s the same procedure as stopping/starting a single server when it comes to the transaction log, any uncommitted transactions will be rollbacked etc. If the failure was really critical the sql server service might not be able to start on the second node either but it will try.
Argyle, You understood the question correct is that the same in active/active
scenario .
Yes it’s same for active/active. Only differences is that you know have one virtual server instance on each node running. If the instance on node 1 fails it will try and start on node 2 and you would have two instances running on node 2. /Argyle
Argyle,
I have a question in the active/active scenario
node1—active
node2–active If node1 fails it failsover to node2 then the node will be running two sqlservers
one of it’s own and the other the failed over sqlserver
am i right.. Thanks,
Yes you are correct. So when running active/active it’s important to make sure that each node has enough resources to support both instances in case of a failover. If node 1 is running with say 2 GB RAM (and using it all) and node 2’s sql instance fails over to it, then the failed over instance will have very little memory to use on node 1 and it will of course have an impact on performance. Same if you have 1 CPU’s in the nodes and each single instance is using say 70% of the CPU. During a failover where you have both instances on the same node, the CPU will not be able to support both instances (combined 140%) without a performance hit. That’s why active/passive scenarios often are more common. You know you have enough resources in each node in case of a failover and you have half the licensing cost (if you use processor licensing). /Argyle
]]>