| Author |
Creating a Bottleneck.
|
|
| SQLCatz 2006-10-24, 6:32 pm |
| I’m using Benchmark factory (TPC-C) on a server that has 3 instances of SQL
Server. Each instance has between 1-3 databases on it. Have set min and max
server memory to 512MB.
What I want:
When the TPC-C transactions are executed I want them to create a memory
and a disk I/O bottleneck.
Apart from the above memory setting what else can I change so that it will
result in a I/O bottleneck?
Cheers!
sqlcatz
| |
| Linchi Shea 2006-10-24, 6:32 pm |
| If the SQL instance only has 512MB, you almost always get a memory and/or I/O
bottleneck with TPC-C when you configure the number of warehouses to be
larger than 10 and include sufficient number of users. If you want to
increase the I/O load, you can reduce the ratios of the read-only
transactions (OrderStatus and StockLevel) in the transaction mix. If you
don't have a really fast I/O subsystem that has a lot of cache, you should
see I/O bottleneck as long as the ffective size of the test database is
significantly larger than the memory size.
Linchi
"SQLCatz" wrote:
> I’m using Benchmark factory (TPC-C) on a server that has 3 instances of SQL
> Server. Each instance has between 1-3 databases on it. Have set min and max
> server memory to 512MB.
> What I want:
> When the TPC-C transactions are executed I want them to create a memory
> and a disk I/O bottleneck.
>
> Apart from the above memory setting what else can I change so that it will
> result in a I/O bottleneck?
>
> Cheers!
> sqlcatz
>
| |
| SQLCatz 2006-10-24, 6:33 pm |
| Hello Linchi!
Thank you for the suggestion.
I tried it.
Apart from getting memory related errors (expected) - and lots of deadlocks
I dont get anything else. On checking the AvgDiskQueueLength - it's just
2.73. I was expecting a much higher value. What else can I do to get the
bottleneck?
Cheers!
sqlcatz
|
|
|
|