dbTalk Databases Forums  

Writting aggregationsisveryslow

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


Discuss Writting aggregationsisveryslow in the microsoft.public.sqlserver.olap forum.



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

Default Writting aggregationsisveryslow - 05-02-2006 , 11:22 AM






Hi

I have a cube that procces 30 millions rows, the query execute very fast
and when write the data to the partitions is relatively fast, however when
the process is writting aggregations and indexs, it take a lot of time, I
tried to see what is consuming Memory or Procesator but I didn't get anything
wrong with this, somebody knows how can I solve this problem

Pd: Sorry by my english
--
Victor Hugo Pozo
MCAD Net

Reply With Quote
  #2  
Old   
Jéjé
 
Posts: n/a

Default Re: Writting aggregationsisveryslow - 05-02-2006 , 01:10 PM






do you use 2005 or 2000?
does the CPU is at 100%?
what is your CPU & memory?


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

Quote:
Hi

I have a cube that procces 30 millions rows, the query execute very fast
and when write the data to the partitions is relatively fast, however
when
the process is writting aggregations and indexs, it take a lot of time, I
tried to see what is consuming Memory or Procesator but I didn't get
anything
wrong with this, somebody knows how can I solve this problem

Pd: Sorry by my english
--
Victor Hugo Pozo
MCAD Net



Reply With Quote
  #3  
Old   
Hugo
 
Posts: n/a

Default Re: Writting aggregationsisveryslow - 05-02-2006 , 02:34 PM



I use AS 2000
No, not even close
Pentium 4- 3Ghz and 2GB Ram
--
Victor Hugo Pozo
MCAD Net


"Jéjé" wrote:

Quote:
do you use 2005 or 2000?
does the CPU is at 100%?
what is your CPU & memory?


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:CEA6E6B8-005E-4F81-BA68-2A439D6072C6 (AT) microsoft (DOT) com...
Hi

I have a cube that procces 30 millions rows, the query execute very fast
and when write the data to the partitions is relatively fast, however
when
the process is writting aggregations and indexs, it take a lot of time, I
tried to see what is consuming Memory or Procesator but I didn't get
anything
wrong with this, somebody knows how can I solve this problem

Pd: Sorry by my english
--
Victor Hugo Pozo
MCAD Net




Reply With Quote
  #4  
Old   
Jeje
 
Posts: n/a

Default Re: Writting aggregationsisveryslow - 05-02-2006 , 10:26 PM



and how many time the process takes?
does your cube contains a distinct count measure?

creating aggregations could take a lot of time if there is a lot of
aggregations and a high density in the content.
a parrallel process and partitions could help you.


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

Quote:
I use AS 2000
No, not even close
Pentium 4- 3Ghz and 2GB Ram
--
Victor Hugo Pozo
MCAD Net


"Jéjé" wrote:

do you use 2005 or 2000?
does the CPU is at 100%?
what is your CPU & memory?


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:CEA6E6B8-005E-4F81-BA68-2A439D6072C6 (AT) microsoft (DOT) com...
Hi

I have a cube that procces 30 millions rows, the query execute very
fast
and when write the data to the partitions is relatively fast, however
when
the process is writting aggregations and indexs, it take a lot of time,
I
tried to see what is consuming Memory or Procesator but I didn't get
anything
wrong with this, somebody knows how can I solve this problem

Pd: Sorry by my english
--
Victor Hugo Pozo
MCAD Net






Reply With Quote
  #5  
Old   
Hugo
 
Posts: n/a

Default Re: Writting aggregationsisveryslow - 05-03-2006 , 09:01 AM



It takes almost 3 hours
Yes, one measure is distinct count
There is just 4 mesaure in the cube, but have many dimensions
How can I create a parallel process ?

--
Victor Hugo Pozo
MCAD Net


"Jeje" wrote:

Quote:
and how many time the process takes?
does your cube contains a distinct count measure?

creating aggregations could take a lot of time if there is a lot of
aggregations and a high density in the content.
a parrallel process and partitions could help you.


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:B28D21C9-F1E6-45AC-911A-58C9178A49C9 (AT) microsoft (DOT) com...
I use AS 2000
No, not even close
Pentium 4- 3Ghz and 2GB Ram
--
Victor Hugo Pozo
MCAD Net


"Jéjé" wrote:

do you use 2005 or 2000?
does the CPU is at 100%?
what is your CPU & memory?


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:CEA6E6B8-005E-4F81-BA68-2A439D6072C6 (AT) microsoft (DOT) com...
Hi

I have a cube that procces 30 millions rows, the query execute very
fast
and when write the data to the partitions is relatively fast, however
when
the process is writting aggregations and indexs, it take a lot of time,
I
tried to see what is consuming Memory or Procesator but I didn't get
anything
wrong with this, somebody knows how can I solve this problem

Pd: Sorry by my english
--
Victor Hugo Pozo
MCAD Net







Reply With Quote
  #6  
Old   
Jéjé
 
Posts: n/a

Default Re: Writting aggregationsisveryslow - 05-03-2006 , 11:42 AM



3h appear to be long.
I do a full process of 2 * 12millions of rows in 10 minutes using AS2005.

if you have a distinct count measure, put this one into a dedicated cube.
1 cube for the DCount measure and 1 cube for the other measures (sums, count
etc...)

search on the MS web site for the parrallel process utility.
but if you have only 1 CPU, the advantage of this utility is low, there is
an improvement, but a small one.



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

Quote:
It takes almost 3 hours
Yes, one measure is distinct count
There is just 4 mesaure in the cube, but have many dimensions
How can I create a parallel process ?

--
Victor Hugo Pozo
MCAD Net


"Jeje" wrote:

and how many time the process takes?
does your cube contains a distinct count measure?

creating aggregations could take a lot of time if there is a lot of
aggregations and a high density in the content.
a parrallel process and partitions could help you.


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:B28D21C9-F1E6-45AC-911A-58C9178A49C9 (AT) microsoft (DOT) com...
I use AS 2000
No, not even close
Pentium 4- 3Ghz and 2GB Ram
--
Victor Hugo Pozo
MCAD Net


"Jéjé" wrote:

do you use 2005 or 2000?
does the CPU is at 100%?
what is your CPU & memory?


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:CEA6E6B8-005E-4F81-BA68-2A439D6072C6 (AT) microsoft (DOT) com...
Hi

I have a cube that procces 30 millions rows, the query execute very
fast
and when write the data to the partitions is relatively fast,
however
when
the process is writting aggregations and indexs, it take a lot of
time,
I
tried to see what is consuming Memory or Procesator but I didn't get
anything
wrong with this, somebody knows how can I solve this problem

Pd: Sorry by my english
--
Victor Hugo Pozo
MCAD Net









Reply With Quote
  #7  
Old   
Hugo
 
Posts: n/a

Default Re: Writting aggregationsisveryslow - 05-11-2006 , 09:30 AM



What I did?
I followed you suggest about create a new partition, in fact I divide the
fact table in 2 tables and make them sources for each partition
The other thing that I had to do is reduce the number of aggregations,
because the first aggregations save faster and the last aggregations take
a lot time for save I still dont know why
Anyway this reduced my time to 40 min, even with one proccesor

Thanks Jéjé for your help

--
Victor Hugo Pozo
MCAD Net


"Jéjé" wrote:

Quote:
3h appear to be long.
I do a full process of 2 * 12millions of rows in 10 minutes using AS2005.

if you have a distinct count measure, put this one into a dedicated cube.
1 cube for the DCount measure and 1 cube for the other measures (sums, count
etc...)

search on the MS web site for the parrallel process utility.
but if you have only 1 CPU, the advantage of this utility is low, there is
an improvement, but a small one.



"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:283345EC-A81D-41A4-8EB9-B90FA4C70F4A (AT) microsoft (DOT) com...
It takes almost 3 hours
Yes, one measure is distinct count
There is just 4 mesaure in the cube, but have many dimensions
How can I create a parallel process ?

--
Victor Hugo Pozo
MCAD Net


"Jeje" wrote:

and how many time the process takes?
does your cube contains a distinct count measure?

creating aggregations could take a lot of time if there is a lot of
aggregations and a high density in the content.
a parrallel process and partitions could help you.


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:B28D21C9-F1E6-45AC-911A-58C9178A49C9 (AT) microsoft (DOT) com...
I use AS 2000
No, not even close
Pentium 4- 3Ghz and 2GB Ram
--
Victor Hugo Pozo
MCAD Net


"Jéjé" wrote:

do you use 2005 or 2000?
does the CPU is at 100%?
what is your CPU & memory?


"Hugo" <Hugo (AT) discussions (DOT) microsoft.com> wrote in message
news:CEA6E6B8-005E-4F81-BA68-2A439D6072C6 (AT) microsoft (DOT) com...
Hi

I have a cube that procces 30 millions rows, the query execute very
fast
and when write the data to the partitions is relatively fast,
however
when
the process is writting aggregations and indexs, it take a lot of
time,
I
tried to see what is consuming Memory or Procesator but I didn't get
anything
wrong with this, somebody knows how can I solve this problem

Pd: Sorry by my english
--
Victor Hugo Pozo
MCAD Net










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.