dbTalk Databases Forums  

Server/Client Processing

microsoft.public.sqlserver.olap microsoft.public.sqlserver.olap


Discuss Server/Client Processing in the microsoft.public.sqlserver.olap forum.



Reply
 
Thread Tools Display Modes
  #1  
Old   
chris harrison
 
Posts: n/a

Default Server/Client Processing - 10-24-2005 , 05:10 AM







Hi there,

We have a AS2000 server (SP4) and a C# application which uses DSO to
connect to the server from a remote client machine (also with SP4, but
only Analysis Manager). Having populated the mart this client app then
makes the call to process the database.

We're seeing the client machine hit 100% CPU and apparently do all of
the work, with the server (the much *bigger*, more capable server) do
nothing beyond a couple of network traffics spikes (presumably squirting
the data across and then receiving the aggregations).

I've been reading around and while there are references to the Execution
Location in the OLAP data provider connection string, it doesn't seem to
be an option in the DSO library?

Is there a way for the client to spawn the processing explicitly on the
server - or do we have to run a small listening app on the server which
we can then have the client call so as to do the processing on the
bigger box?

We have seen other applications (for example a 'toy' one using the
FoodMart example) which seem to do a lot more work on the server. These
were tested (by tech partners) but on boxes with SP3a. Is the SP a
possible difference or are there potential characteristics of the cube
which determine this behavioural delta?

cheers for any suggestions,

- chris.


Reply With Quote
  #2  
Old   
Denny Lee
 
Posts: n/a

Default Re: Server/Client Processing - 10-24-2005 , 11:59 AM






I do not recall the reason why your client box is getting pegged more than
the server box, but I have definitely seen this behaviour for quite awhile.
Saying this, the best option is as you had suggested which is to have the
Olap server call the app on its own so that way all the pegging is on the
Olap server (vs. the client server). By the way, where is your SQL server
that contains all of the fact data that the OLAP server is going to process
against? For example, if your SQL data is on the client machine (as per
below in this case) vs. being on the OLAP server, that would explain your
CPU pegging. When OLAP processes, for awhile, the server that contains the
SQL data is the one that is pegged during processing. One the partitions
and aggregates are actually being created, then the OLAP server will get the
bulk of the processing.

--
HTH!
Denny Lee
<dennyglee_at_hotmail_dot_com>
space: http://spaces.msn.com/members/denster/



"chris harrison" <unsliced (AT) gmail (DOT) com> wrote

Quote:
Hi there,

We have a AS2000 server (SP4) and a C# application which uses DSO to
connect to the server from a remote client machine (also with SP4, but
only Analysis Manager). Having populated the mart this client app then
makes the call to process the database.

We're seeing the client machine hit 100% CPU and apparently do all of
the work, with the server (the much *bigger*, more capable server) do
nothing beyond a couple of network traffics spikes (presumably squirting
the data across and then receiving the aggregations).

I've been reading around and while there are references to the Execution
Location in the OLAP data provider connection string, it doesn't seem to
be an option in the DSO library?

Is there a way for the client to spawn the processing explicitly on the
server - or do we have to run a small listening app on the server which
we can then have the client call so as to do the processing on the
bigger box?

We have seen other applications (for example a 'toy' one using the
FoodMart example) which seem to do a lot more work on the server. These
were tested (by tech partners) but on boxes with SP3a. Is the SP a
possible difference or are there potential characteristics of the cube
which determine this behavioural delta?

cheers for any suggestions,

- chris.




Reply With Quote
  #3  
Old   
Dave Wickert [MSFT]
 
Posts: n/a

Default Re: Server/Client Processing - 10-25-2005 , 06:46 PM



I have written many DSO apps and except for receiving the progress reports
coming back to the server you should NOT be seeing the client be the
bottleneck. It might run at high CPU but I guarantee you, it is NOT doing
the real work of processing.

I would enable the system-wide processing log file. Bring up AM, right-click
on the server and select Properties. On the "Logging" tab, at the bottom,
enable the processing log file and enter something like C:\Processing.txt as
the file. Then anytiime anyone does any processing, you will see the same
progress message you see interactively in that file.
--
Dave Wickert [MSFT]
dwickert (AT) online (DOT) microsoft.com
Program Manager
BI Systems Team
SQL BI Product Unit (Analysis Services)
--
This posting is provided "AS IS" with no warranties, and confers no rights.


"chris harrison" <unsliced (AT) gmail (DOT) com> wrote

Quote:
Hi there,

We have a AS2000 server (SP4) and a C# application which uses DSO to
connect to the server from a remote client machine (also with SP4, but
only Analysis Manager). Having populated the mart this client app then
makes the call to process the database.

We're seeing the client machine hit 100% CPU and apparently do all of
the work, with the server (the much *bigger*, more capable server) do
nothing beyond a couple of network traffics spikes (presumably squirting
the data across and then receiving the aggregations).

I've been reading around and while there are references to the Execution
Location in the OLAP data provider connection string, it doesn't seem to
be an option in the DSO library?

Is there a way for the client to spawn the processing explicitly on the
server - or do we have to run a small listening app on the server which
we can then have the client call so as to do the processing on the
bigger box?

We have seen other applications (for example a 'toy' one using the
FoodMart example) which seem to do a lot more work on the server. These
were tested (by tech partners) but on boxes with SP3a. Is the SP a
possible difference or are there potential characteristics of the cube
which determine this behavioural delta?

cheers for any suggestions,

- chris.




Reply With Quote
Reply




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off



Powered by vBulletin Version 3.5.3
Copyright ©2000 - 2012, Jelsoft Enterprises Ltd.