Re: MSCS Questions
From: Michael C# (xyz_at_abcdef.com)
Date: 02/19/05
- Next message: JTC ^..^: "Re: Scripting a SQL database"
- Previous message: Robbe Morris [C# MVP]: "Re: Scripting a SQL database"
- In reply to: Andrew J. Kelly: "Re: MSCS Questions"
- Next in thread: Andrew J. Kelly: "Re: MSCS Questions"
- Reply: Andrew J. Kelly: "Re: MSCS Questions"
- Messages sorted by: [ date ] [ thread ]
Date: Sat, 19 Feb 2005 18:46:44 -0500
I had 2 GB before, and now have 4 GB with the /3GB switch. I'm not using
AWE or PAE or any of that good stuff. I wouldn't think 4 GB of RAM is an
excessive amount that would degrade performance. We're not talking about
the overhead of AWE. The SQL Server is set to dynamically use up to 3072 MB
max.
I've run traces against memory, network, disk and processor counters
including:
Processor : %Processor Time;
System : Processor Queue Length;
System : %Total Processor Time;
PhysicalDisk : Avg. Disk Queue Length;
Physical Disk : % Disk Time counter;
Memory : Pages/Sec;
Memory : Available Bytes;
SQL Server Locks : Average Wait Time (ms);
SQLServer : SQL Statistics : Batch Requests/Sec;
SQL Server Buffer Manager : Buffer Cache Hit Ratio;
SQL Server Access Methods Object : Full Scans/sec
Among several other counters, including network I/O counters and thread
counters. All the traces I've run have favorable results, and they all
point to no bottlenecks in the system.
Do you have any recommendations for more traces I can run that will help me
find out where the issues are so I can address them accordingly?
Thanks,
Michael C., MCDBA
"Andrew J. Kelly" <sqlmvpnooospam@shadhawk.com> wrote in message
news:uR4ZtmsFFHA.3596@TK2MSFTNGP12.phx.gbl...
> There is little effect on performance one way or the other with clustering
> itself. While faster processors may make a difference in performance more
> memory may not. Again it all depends on how well it was utilized before
> and now. If Sql Server can not take advantage of more memory then having
> too much can actually produce more overhead than you need. You need to
> run some traces to find out where the issues are and address them
> accordingly.
>
> --
> Andrew J. Kelly SQL MVP
>
>
> "Michael C#" <xyz@abcdef.com> wrote in message
> news:asKRd.1224$9e6.957@fe12.lga...
>> Well - the database is as optimized as we can make it (AFAIK), and we
>> just re-wrote some portions of the .NET code we're using to populate it
>> to increase performance. In theory, we're dealing with a much
>> higher-powered server solution than we had, including more memory and
>> faster processors.
>>
>> I understand that clustering isn't a performance benefit, but was
>> wondering if it might potentially be hurting performance if the
>> configuration is not 'tweaked' in some fashion.
>>
>> Thx
>>
>> "Andrew J. Kelly" <sqlmvpnooospam@shadhawk.com> wrote in message
>> news:uXrWlfpFFHA.2676@TK2MSFTNGP12.phx.gbl...
>>> If you don't have bottlenecks in the counters it may simply be that the
>>> code itself is not as efficient as it might be. If you were not pushing
>>> the hardware before to the point that you had bottlenecks there is no
>>> reason to assume it will be faster elsewhere. Clustering does nothing
>>> for performance it is a hardware failover solution only. The SAN will
>>> only help with performance if you have I/O issues before.
>>>
>>> --
>>> Andrew J. Kelly SQL MVP
>>>
>>>
>>> "Michael C#" <xyz@abcdef.com> wrote in message
>>> news:nxzRd.106$5c6.70@fe12.lga...
>>>> Hi all,
>>>>
>>>> I've just completed setting up two Win 2003 Servers w/ SQL Server 2K
>>>> (SP3a) in an MSCS Cluster. Everything is working, but the performance
>>>> is not as good as I thought it would be. Here's a quick summary:
>>>>
>>>> -Each server has 4 GB of RAM, gigabit ethernet and plenty of RAID 1
>>>> storage both locally and on the SAN.
>>>> -I've run System Monitor and found no bottlenecks in the system (Hard
>>>> Drive, Network, Memory, or CPU).
>>>> -I've set up separate filegroups for data and non-clustered indexes on
>>>> different drives and put the transaction logs on their own drive.
>>>> -I've run SQL Best Practices Analyzer and received rave reviews for the
>>>> configuration.
>>>> -I've optimized the database as much as possible, but the performance
>>>> seems to be roughly equivalent to the performance on an old testing
>>>> server we had been using up to this point.
>>>> -I added the /3GB switch and set SQL to dynamically use up to a maximum
>>>> of 3,072 MB.
>>>>
>>>> I was just wondering if there are some configuration tweaks I'm missing
>>>> that can be made to SQL Server 2K on the Cluster to enhance
>>>> performance?
>>>>
>>>> TIA
>>>>
>>>
>>>
>>
>>
>
>
- Next message: JTC ^..^: "Re: Scripting a SQL database"
- Previous message: Robbe Morris [C# MVP]: "Re: Scripting a SQL database"
- In reply to: Andrew J. Kelly: "Re: MSCS Questions"
- Next in thread: Andrew J. Kelly: "Re: MSCS Questions"
- Reply: Andrew J. Kelly: "Re: MSCS Questions"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
