How Man Partitions

How Man Partitions

Post by news.btinternet.co » Mon, 02 Jul 2001 16:11:24



We have a cube which consists of data from about 180 feeds gathered each
day. We have a cube which has just a single day's data in that covers about
95% of reports that we need. However, on an average day we may need to
reload about 5 feeds due to calculation errors in the source data. We are
thinking that we should have 1 partition per feed and set the slice and
filter values accordingly from VB and then do a full process on just that
partition whenever data arrives

Is it OK to have 180 partitions

 
 
 

How Man Partitions

Post by Wayne Robertso » Wed, 04 Jul 2001 02:41:04


There is no defined limit on the number of partitions you can have in a
cube.  As long as you have the Enterprise Edition of Analysis Services you
can create as many partitions as you need.  Just keep in mind that managing
that many partitions may be difficult from an administrative standpoint.


Quote:> We have a cube which consists of data from about 180 feeds gathered each
> day. We have a cube which has just a single day's data in that covers
about
> 95% of reports that we need. However, on an average day we may need to
> reload about 5 feeds due to calculation errors in the source data. We are
> thinking that we should have 1 partition per feed and set the slice and
> filter values accordingly from VB and then do a full process on just that
> partition whenever data arrives

> Is it OK to have 180 partitions


 
 
 

How Man Partitions

Post by Doug Welz » Thu, 05 Jul 2001 05:20:01


Another option to look at is maintaining a set of "today" partitions
and one containing the rest of your history.  After the day closes
(and you know the data is good), merge your 180 partitions into the
larger partition.  This might make maintenance easier.

> We have a cube which consists of data from about 180 feeds gathered each
> day. We have a cube which has just a single day's data in that covers about
> 95% of reports that we need. However, on an average day we may need to
> reload about 5 feeds due to calculation errors in the source data. We are
> thinking that we should have 1 partition per feed and set the slice and
> filter values accordingly from VB and then do a full process on just that
> partition whenever data arrives

> Is it OK to have 180 partitions

 
 
 

1. Mulit partition ESE across a MAN?

Hello all

I wonder what the opinion here is about running a partitioned DB for
data warehousing across a MAN - are there any rules of thumb or
thoughts on this?

We want to split the data nodes across sites for high availability
reasons, but are a bit worried about the performance side effects.
Bandwidth is good, RTT is also not bad - should I worry about the
latency above all else, or are there other issues?

This is for new data warehouses, as ever there is no chance to test
properly any usage patterns - one day they'll go live, next day I
still want to have a job.

Running AIX 5.2, HACMP, v8.1 fp1.

thanks for any thoughts on this.

2. Call DLL's via a Trigger ?

3. Man oh Man - Need VB Help for Using Access Databases

4. Finding unique field entries

5. You're the man, man !!!

6. SQL Mail

7. Urgent! Process partition fails when partition is cloned

8. CommanDLine -creating partitions - processing partitions

9. Partitioned views - how to size partitions?

10. CommanDLine -creating partitions - processing partitions

11. Convert a non-partitioned table to partitioned table

12. partition views vs partition tables