Click here to monitor SSC
SQLServerCentral is supported by Red Gate Software Ltd.
 
Log in  ::  Register  ::  Not logged in
 
 
 

Managing MaxMemory Values on an Active-Active Two Node Cluster

By Michael Lacy,

Okay, you have your brand new, two node, Windows Cluster built, and SQL Server installed and running on an active-active cluster. When you bought the hardware, you loaded-up on memory with each node having 256GB of the fastest memory money can buy. You’ve configured your two instances with a max memory value of 245GB and everything runs great - until one day one of the nodes has an error that places both instances onto one node. Now you have two instances, each with a max memory value of 245GB running on one piece of hardware with 256GB of memory. 245GB + 245GB is more than 256GB. Ouch! You could take the easy way out and make sure that the max memory on each instance adds-up to less than 256GB, but that means that when everything is working well, you’re wasting about half of the available memory.

How do you maximize the use of your installed memory, and yet automatically reconfigure your max memory value settings when an unexpected fail-over occurs? You just need to link your two SQL Server instances, create some empty text files, create a few of SQL Server Agent jobs, and you’re good-to-go.

My Windows Cluster has been correctly built and drive dependencies are correctly defined for my two SQL Server instances. Here’s my server hardware setup:

  • Node1=MyServer1: Should host my production instance which is named: ProdInst\ProdDBs and has disk E: (a SAN attached disk)
  • Node2=MyServer2: Should host my reporting instance which is named: ReptInst\ReptDBs and has disk H: (a SAN attached disk)

(I have a C: drive and other SAN drives for my T-logs and TempDB, but this process does not reference those drives.)

First, link the two SQL Server instances. Edit and run this code on each SQL Server, linking both instances to one another. This was run on my production instance to link to my reporting instance:

USE [master]
GO
EXEC master.dbo.sp_addlinkedserver @server = N'ReptInst\ReptDBs', @srvproduct=N'SQL Server'

This was run on my reporting instance to link to my production instance:

USE [master]
GO
EXEC master.dbo.sp_addlinkedserver @server = N'ProdInst\ProdDBs', @srvproduct=N'SQL Server'

Note: RPC should be set to “True” on each linked server.

Next, create an empty text file on each node. Create these on your SAN disks that will fail-over per you Windows Cluster Manager configuration. I’ll call these my “SAN Test Files”.

  • On SAN Disk – for ProdInst\ProdDBs: E:\DBA\ProdInst.txt
  • On SAN Disk – for ReptInst\ReptDBs: H:\DBA\ReptInst.txt

Then, create two SQL Server Agent jobs on each SQL Server instance. One job to set the max memory value to what you want it to be when just one instance is on the node, and a second SQL Server Agent job to set the max memory value to what you’d like it to be if both SQL Server instances are running on the same node. If just one instance is on a node, I want the max memory value set to 245GB, and if both instances are running on the same node, I want the max memory value for my ProdInst\ProdDBs instance to be 173GB and my ReptInst\ReptDBs to be 72GB.

Memory is usually expressed in MBs, not GBs, so I use the following values:

  • 245GB = 250880MB
  • 173GB = 177152MB
  • 72GB = 73728MB

The jobs to set the max memory values are named: 'Set MaxMemory Value on PRODDBS to 245GB' and 'Set MaxMemory Value on REPTDBS to 245GB'. Here is the code:

EXEC sys.sp_configure N'show advanced options', N'1'  RECONFIGURE WITH OVERRIDE
GO
EXEC sys.sp_configure N'max server memory (MB)', N'250880'
GO
RECONFIGURE WITH OVERRIDE
GO
EXEC sys.sp_configure N'show advanced options', N'0'  RECONFIGURE WITH OVERRIDE
GO

Here is the second job, named: 'Set MaxMemory Value on PRODDBS to 173GB':

EXEC sys.sp_configure N'show advanced options', N'1'  RECONFIGURE WITH OVERRIDE
GO
EXEC sys.sp_configure N'max server memory (MB)', N'177152'
GO
RECONFIGURE WITH OVERRIDE
GO
EXEC sys.sp_configure N'show advanced options', N'0'  RECONFIGURE WITH OVERRIDE
GO

I have a third job named: 'Set MaxMemory Value on REPTDBS to 72GB':

EXEC sys.sp_configure N'show advanced options', N'1'  RECONFIGURE WITH OVERRIDE
GO
EXEC sys.sp_configure N'max server memory (MB)', N'73728'
GO
RECONFIGURE WITH OVERRIDE
GO
EXEC sys.sp_configure N'show advanced options', N'0'  RECONFIGURE WITH OVERRIDE
GO

I need one more SQL Server Agent job. I’ll call this job: “Alert and Set MaxMemory on Restart”. It exists on each SQL Server instance and is scheduled to “Start automatically whenever the SQL Server Agent starts”. It will perform 2 job steps:

  1. Send an email with the instance name, and the node that it is running on.
  2. Set the max memory values appropriately if both instances are running on the same node, or independently with just one instance on each node.

“Alert and set MaxMemory on Restart” job steps:

Here are the details of this last job.

Job step 1: (This code is basically the same on either instance, but must be edited for you database mail values) (see edit comments). This step sends an email with the instance name, and the node that it is running on.

-- Send an email with the instance name, and the node that it is running on.
USE master
DECLARE @importance AS VARCHAR(6)
DECLARE @body AS NVARCHAR(1000)
DECLARE @Subject  AS NVARCHAR(1000)
DECLARE @InstanceName AS VARCHAR(100)DECLARE @NodeName AS NVARCHAR(100)
DECLARE @recipientsList VARCHAR(100)SELECT  @recipientsList ='myemail@myworkplace.com' -- Edit
SELECT  @InstanceName =@@SERVERNAME
SELECT  @NodeName = CAST(SERVERPROPERTY('ComputerNamePhysicalNetBIOS') AS NVARCHAR(200))
SELECT  @Subject  = '!!! ALERT !!! - Fail over happened for instance: '+@InstanceName
SELECT  @body = 'Fail over happened for instance: '+@InstanceName + '. This instance is currently running on node: '+@NodeName
SELECT  @importance ='High'
EXEC msdb.dbo.sp_send_dbmail
     @profile_name = 'ProdDBs Profile',  -- Edit for your database mail profile
     @recipients = 'myemail@myworkplace.com',  -- Edit for your email operator profile
     @subject = @subject ,
     @body = @body,
     @body_format = 'HTML' ,
     @importance=@importance

Job step 2 for my ProdInst\ProdDBs instance: This sets the appropriate MaxMemory values. If my ProdInst\ProdDBs and my ReptInst\ReptDBs are both on the same server then set ProdInst\ProdDBs to use 173GB of memory and set ReptInst\ReptDBs to use 72GB of memory.

-- Set appropriate MaxMemory values. If my ProdInst\ProdDBs and my ReptInst\ReptDBs are both on the 
-- same server then set ProdInst\ProdDBs to use 173GB of memory and set ReptInst\ReptDBs to use 
-- 72GB of memory.
  SET NOCOUNT ON
DECLARE @FileName varchar(255)
DECLARE @File_Exists int
/* Edit below as necessary. You are testing to see if two instances are running on one node. So for my ProdInst\ProdDBs instance I want to know if the H: drive has been failed over and is running on this one node. See the “SAN Test Files” */
SELECT @FileName='H:\DBA\ReptInst.txt' EXEC Master.dbo.xp_fileexist @FileName, @File_Exists OUT
IF @File_Exists = 1
BEGIN
EXEC sys.sp_configure N'show advanced options', N'1'  RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'max server memory (MB)', N'177152'
RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'show advanced options', N'0'  RECONFIGURE WITH OVERRIDE
EXEC [REPTINST\REPTDBS].msdb.dbo.sp_start_job @job_name = 'Set MaxMemory Value on REPTDBS to 72GB'
END
ELSE
BEGIN
PRINT 'PRODINST\PRODDBS and REPTINST\REPTDBS are NOT running on the same node'
EXEC sys.sp_configure N'show advanced options', N'1'  RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'max server memory (MB)', N'250880'
RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'show advanced options', N'0'  RECONFIGURE WITH OVERRIDE
EXEC [REPTINST\REPTDBS].msdb.dbo.sp_start_job @job_name = 'Set MaxMemory Value on REPTDBS to 245GB'
END

Job step 2 for my ReptInst\ReptDBs instance: This job sets memory values.

-- Set appropriate MaxMemory values. If my ProdInst\ProdDBs and my ReptInst\ReptDBs are both on the 
-- same node then set ProdInst\ProdDBs to 173GB of memory and set ReptInst\ReptDBs to use 72GB 
-- of memory.
SET NOCOUNT ON
DECLARE @FileName varchar(255)
DECLARE @File_Exists int
/* Edit below as necessary. You are testing to see if two instances are running on one node. So for my ReptInst\ReptDBs instance I would want to know if the E: drive has been failed over and is running on this one node. See the “SAN Test Files” */
SELECT @FileName='E:\DBA\ProdInst.txt' EXEC Master.dbo.xp_fileexist @FileName, @File_Exists OUT  
IF @File_Exists = 1
BEGIN
EXEC sys.sp_configure N'show advanced options', N'1'  RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'max server memory (MB)', N'73728'
RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'show advanced options', N'0'  RECONFIGURE WITH OVERRIDE
EXEC [PRODINST\ProdDBs].msdb.dbo.sp_start_job @job_name = 'Set MaxMemory Value on PRODDBS to 173GB'
END
ELSE
BEGIN
PRINT 'REPTINST\REPTDBS and PRODINST\PRODSDBS are NOT running on the same node'
EXEC sys.sp_configure N'show advanced options', N'1'  RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'max server memory (MB)', N'250880'
RECONFIGURE WITH OVERRIDE
EXEC sys.sp_configure N'show advanced options', N'0'  RECONFIGURE WITH OVERRIDE
EXEC [PRODINST\PRODSDBS].msdb.dbo.sp_start_job @job_name = 'Set MaxMemory Value on PRODDBS to 245GB'
END

That’s it. Now your max memory values should always be correctly set, to allow you to use the maximum amount of memory, whether they’re running alone on a node, or if both instances are running on the same node.

Total article views: 3561 | Views in the last 30 days: 26
 
Related Articles
FORUM

how to configure sql server memory

how to configure sql server memory

FORUM

sp_configure - reconfigure

sp_configure - reconfigure

FORUM

Remote RECONFIGURE

Need ability to run RECONFIGURE across numerous remote servers

FORUM
FORUM

Configure Physical Memory

How to configure physical memory in sqlserver 2005

 
Contribute

Join the most active online SQL Server Community

SQL knowledge, delivered daily, free:

Email address:  

You make SSC a better place

As a member of SQLServerCentral, you get free access to loads of fresh content: thousands of articles and SQL scripts, a library of free eBooks, a weekly database news roundup, a great Q & A platform… And it’s our huge, buzzing community of SQL Server Professionals that makes it such a success.

Join us!

Steve Jones
Editor, SQLServerCentral.com

Already a member? Jump in:

Email address:   Password:   Remember me: Forgotten your password?
Steve Jones