Does anyone have experience with migration to a 64 bit system and increasing
RAM accessible to Analysis Services beyond the 3GB limit with respect to
improved query performance? If so, does this enable more of an OLAP database
to be loaded into RAM possibly by running post-processing queries to warm the
server cache and deliver better query response?

We have an application that delivers reports using Active Server Pages that
retrieve data from MS Analysis Services 2K cubes. At peak times, query
response times increase to 40 seconds from seconds at quieter times. Our main
cube is 900 MB on disk, has 5 physical dimensions and around 30 measures. The
largest dim is a parent-child with 80kmembers. We believed that aggregation
design is reasonably set for the type of queries that are generated.

Before investing in the additional harware/software we would like to receive
any comments/feedback from anyone who has carried out this type of upgrade to
their Analysis Services platform.

TIA

P Taylor

Some interesting material here:
http://www.microsoft.com/technet/pro.../64btdwc4.mspx
It is a larger benchmark with memory utilization on a 64-bit system.

However, before you go down that path I'd be more concerned that you were
running out of memory on the web server. It is not uncommon for PTS to take
considerable amount of memory in this type of situation. See what happens if
you take some of your application and run it in a 2nd and a 3rd instance in
different virtual directories with settings so that each VB runs in its own
process. This would allow PTS to cover more memory on the middle tier.
--
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.

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

Are you completely sure you've built the aggregations you need? If you have
an 80K member parent-child dimension, unless you've built some aggregations
manually you may not be hitting any aggregations if you're querying below the
All Member on that dimension. Take a look at the following article on my blog
to see how to find out if you are:
http://spaces.msn.com/members/cwebbbi/Blog/cns!1pi7ETChsJ1un_2s41jm9Iyg!119.entry

My first bit of advice to improve performance would be to remodel this
parent-child dimension as a regular dimension, if you can - apart from the
fact that it makes aggregation design much easier and efficient, there are
other performance benefits too. If you can't remodel it, building
aggregations manually might be useful (although I recommend it as a last
resort).

Regards,

Chris

--
Blog at:
http://spaces.msn.com/members/cwebbbi/


"ptaylor" wrote: