Two pointers:
http://www.microsoft.com/technet/pro.../64btdwc4.mspx
http://www.microsoft.com/technet/pro.../ansvcs64.mspx
as some information.
Going to 64-bit is important for capacity purposes; not for just performance
per-say.
For # of parallel processing requests, I would contact Sanjay Soni at
Unisys. [Sanjay.Soni (AT) unisys (DOT) com]
He has done presentations at various conferences concerning the work he did
on large 64-bit systems. 10 in parallel might be as much as you can go
before you start to see diminishing returns. It really depends on what
relational work needs to be done with each. Running much more than 10 at the
same time means that other bottlenecks start to appear, e.g. the rate of
memory allocation in the msmdsrv starts to kick in some algorithms which
shifts processing to start using temp files even though physical memory
exists. But I will let Sanjay share his experiences.
Hope that helps.
--
Dave Wickert [MSFT]
dwickert (AT) online (DOT) microsoft.com
Program Manager
BI SystemsTeam
SQL BI Product Unit (Analysis Services)
--
This posting is provided "AS IS" with no warranties, and confers no rights.
"Prasad" <Prasad (AT) discussions (DOT) microsoft.com> wrote
Quote:
Hi,
Does any body have real bench marks made on MSAS 2000 on 64 bit.
Data Load times , Aggregation processing times, How many dimensions in the
cube, typical big dimension sizes, fact table row counts etc.
I am also interested in how many parllel processes any body has run? We
are
looking at running atleast 10 agents in parllel.
I will appreciate if you can provide the hardware config along with the
bench marks.
- Amount of RAM
- Processer power
- Virtual memory
- Typical performance boosting settings on the Analysis server etc
Thanks for the help. |