![]() | |
![]() |
| | Thread Tools | Display Modes |
#1
| |||
| |||
|
#2
| |||
| |||
|
|
Hi, I would like to know the following with an expectation of good performance... - Max. number of records a cube can hold - The max. size in the disk for that - Data loading time (initial, and regular updates) - Performance with such a huge amount of data - With medium level of complexity in querying and reporting For example, what about having nearly 5 million records that occupies few GB of disk space? And the performance for this? Thankx Philipus -- Message posted via http://www.sqlmonster.com |
#3
| |||
| |||
|
|
5 millions is somewhat small. the hardware is the most important part. a server with 2 proc. 4Gb load 20 millions of rows in 20minutes. Why this time? because I have a lot of dimensions (more then 20)!!! (also the disks are badly optimized) fewer dimensions result in smaller process time. in the past I have created a cube with 5 millions of rows and 6 dimensions loaded in 2 minutes. look at the Project "REAL" http://www.microsoft.com/technet/pro.../projreal.mspx (there is more then 8billions of rows) or the "old" T3 project (created in 2001) http://www.microsoft.com/sql/evaluat...rabytecube.asp I hope this helps you. "Philipus via SQLMonster.com" <forum (AT) SQLMonster (DOT) com> wrote in message news:50CFB3C9F5CF0 (AT) SQLMonster (DOT) com... Hi, I would like to know the following with an expectation of good performance... - Max. number of records a cube can hold - The max. size in the disk for that - Data loading time (initial, and regular updates) - Performance with such a huge amount of data - With medium level of complexity in querying and reporting For example, what about having nearly 5 million records that occupies few GB of disk space? And the performance for this? Thankx Philipus -- Message posted via http://www.sqlmonster.com |
#4
| |||
| |||
|
|
Aggreed - 5 mio is small.... Heard some pointers say that you should consider partitioning when reaching 40-50 mio. rows.... So it shouldn't be a problem. However lots of times it isnt the record count but dimension design trouble that gives you performance trouble. Make hierachies as much as possible - avoid combining large flat dimensions. "Jéjé" <willgart (AT) BBBhotmailAAA (DOT) com> wrote in message news:eG44QgPgFHA.1416 (AT) TK2MSFTNGP09 (DOT) phx.gbl... 5 millions is somewhat small. the hardware is the most important part. a server with 2 proc. 4Gb load 20 millions of rows in 20minutes. Why this time? because I have a lot of dimensions (more then 20)!!! (also the disks are badly optimized) fewer dimensions result in smaller process time. in the past I have created a cube with 5 millions of rows and 6 dimensions loaded in 2 minutes. look at the Project "REAL" http://www.microsoft.com/technet/pro.../projreal.mspx (there is more then 8billions of rows) or the "old" T3 project (created in 2001) http://www.microsoft.com/sql/evaluat...rabytecube.asp I hope this helps you. "Philipus via SQLMonster.com" <forum (AT) SQLMonster (DOT) com> wrote in message news:50CFB3C9F5CF0 (AT) SQLMonster (DOT) com... Hi, I would like to know the following with an expectation of good performance... - Max. number of records a cube can hold - The max. size in the disk for that - Data loading time (initial, and regular updates) - Performance with such a huge amount of data - With medium level of complexity in querying and reporting For example, what about having nearly 5 million records that occupies few GB of disk space? And the performance for this? Thankx Philipus -- Message posted via http://www.sqlmonster.com |
![]() |
| Thread Tools | |
| Display Modes | |
| |