Question:
Hello,
Every morning, we load 46 flat files from 1h00 to 8h00 (size between 300Mo and 2Go) in a R/3 source system.
Then the files are loaded in several huges tables for extraction.
Afterwards, the data are extracted to our BW system.
The Infocube is loaded in parallel, the processing option of the infopackages is " Data Target only".
(We have a deadline and can't spend much time in PSA)
The R/3 tables are loaded and extracted to BW at the same time.
Since the beginning of the this week, we are encoutering many performance problems during the loading.
The extraction to BW is very long, each data-package is sent every ten minutes. Last week , the same data-packages were sent every 30 sec !!!
Most of the extractions job (bi_requ*) end in a short dump :
DBIF_RSQL_SQL_ERROR
Database error text........: "ORA-01555: snapshot too old: rollback segment
number 11 with name "PRS_10" too small?"
Internal call code.........: "[RSQL/FTCH/ZZBWTR30 ]"
In fact, we have 20 rollbacks segments, each of them 400Mo,
which is already a lot , and validated by SAP.
We didn't detect any memory or CPU bottleneck during the night, that
was confirmed by our unix administrators.
Anyone has an idea ?
Any help would be highly appreciated.
Thank you very much.
Oracle 9.2 AIX Version5 SAP 4.6C/SAP Business Information Warehouse 3.10
Answer:
Not sure, but OSS note 60233 might help.
Answer:
Hi
From Database perspective seems to me as there are primarily 3 reasons for these.
Plz refer Note -185822 for these.
Hope it Helps
_________________
Chetan
@ CP...