on 01-26-2009 9:51 PM
On Manage for an InfoCube, then on the Performance tab, there are several things that are allowed to happen with Indexes. I think the norm is to drop and index, load data, then rebuild an index, but not sure what is going on. Is this all internal to BW, or would a backend BW person make changes that affect the index area.
Can you point me to information for this area?
Thanks, Keith J
Hello Keith,
I normally run the program : SAP_INFOCUBE_INDEXES_REPAIR as a background job after any manual loading activity which may have fragmented the index.
This job takes care of all index rebuilding activities.
Regards,
Siddhesh
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Keith you only drop an index if you really have to
The reason for dopping them is two fold:
1. if you have a large load and are creating lots of new DIMids - if you have small number of new DIMs and a large DIM table then it makes no sense dropping the DIM index as you need it othrwise your loads are going to run like dogs. However if you have a large new load with lots of new DIMs then there is no need to waste CPU updating the DIM index at the same time as inserting and you dont need the index to read the DIM table because it so small
2. If you have an oracle backend and are experiencing deadlocks in parallel writes to DIMs. ie if you have 3 parallel infopackages updating the info cube each with new master data in it. And some of that new master data is creating new DIMids in the same oracle block - then deadlock! No workaround and no workround planned. Unfortunately then in all circumstances you have to drop the indexes as a precaution otherwise you are forever going to get called out each night.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
In gemeral, before loading the data to Cube you need to delete the Indexs first then load the data then create Index and DB stats. This BW consultat need to do it. This is not manual thing, but you can also do it Manually. Better keep it in Process chains.
Start
|
Delete Index
|
Load data to Cube
|
Create Index
|
Create DB stats.
In Process chans , in Data Traget Administartion you can find Delete Index , Generate Index and Construct Database Statistics .So us it.
Thanks
Reddy
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi
Generally, When we load Data to the Data Targets we do this activity of Drop Indexes and ReBuild to make sure that we are giving less fetching time while Reporting for the Info Provider, for this activity we may need Administrative Activities Role & Authorization, which is normally assigned to BI Team.
Hope it helps
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Kieth,
You understanding is correct, you have to drop index then load data and recreate the index after the newly inserted records are considered.
This enables faster load performance since the inserts take lesser time, also the query performance in data selection is improved.
Refer to this link below.
http://help.sap.com/saphelp_nw70/helpdata/EN/80/1a6473e07211d2acb80000e829fbfe/frameset.htm
Hope this helps.
Regards,
Sunmit.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
83 | |
10 | |
10 | |
9 | |
7 | |
6 | |
5 | |
5 | |
4 | |
3 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.