Memory in a 2 node Active/Active cluster

#41

The "Lock Pages in Memory" simple means the OS cannot force SQL to
relinquish memory. The OS can ask and SQL can comply on its own schedule
and release memory. However, there is no balancing between instances. All
user applications get the memory panic message, even the newly started
instance.

With larger systems having high core counts like yours, you may have to
reduce memory to 24027 GB of RAM. You can get it a little tighter in
Enterprise Edition SQL because "lock Pages" actually works. "Lock Pages" is
not recognized by Standard Edition SQL Server.

--
Geoff N. Hiten
Principal SQL Infrastructure Consultant
Microsoft SQL Server MVP

"Russ" <Russ (AT) discussions (DOT) microsoft.com> wrote

Quote:

Hi,
I have a scenario involving a two node cluster each running a SQL 2005
(x64)
instance. Each node has 64GB of RAM and 'lock pages in memory' is set.

If I configure, via 'Max Server Memory', each node to use say 50GB of
RAM -
what happens during a failover. I'm guessing that because of the 'lock
pages
in memory' the surviving instance will not release any memory to
accomodate
the failed instance and will be forced to compete for the remaining 14 GB
of
RAM with the OS.

Whats the best practice here? Should I configure each node to use only
30GB
of RAM? (leaving 4GB for the OS).

Thanks for any advice....

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Memory in a 2 node Active/Active cluster

microsoft.public.sqlserver.clustering microsoft.public.sqlserver.clustering

Re: Memory in a 2 node Active/Active cluster - 11-21-2008 , 01:03 PM