Interested to Buy Any Domain ? << Click Here >> for more details...
How to identify bottlenecks in
sources,targets,mappings,workflow,system and how to
increase the performance?
- 4 Answers
- 35630 Views
- I also Faced
- E-Mail Answers
Answers were Sorted based on User's Feedback
Answer / praveenkumar.b
Source:
Create the Filter transformation after all the Source
Qualifiers and make the filter condition FALSE so that the
data will not go beyond this trasformation. Then run the
session and find out the time taken from source. If you
feel there is some lack in performance, then suggest the
necessary index creation in Pre Session.
Note: If the source is File, then there is no possibility
of performance problme in source side
Target:
Delete the target table from the mapping and create the
same structure as a Flat file. Run the session and find out
the time taken to write the file. If you feel problem in
performance, then delete the INDEX of the table before
loading the data. In post Session, Create the same index
Note:If the target is File, then there is no possibility of
performance problme in target side
Mapping:
The below steps need to be consider
#1. Delete all the transformations and make it as single
pass through
#2. Avoid using more number of transformations
#3. If you want to use more filter transformation, then use
Router transformation instead of it
#4. Calculate the index and data cache properly for
Aggregator, Joiner, Ranker, Sorter if the Power center is
lower version. Advance version, Power center itself will
take care of this
#5. Always pass the sorted i/p's to Aggregator
#6. Use incremental aggregation
#7. Dont do complex calculation in Aggregator
transformation.
Session:
Increas the DTM buffer size
System:
#1. Increase the RAM capacity
#2. Avoid paging
| Is This Answer Correct ? | 27 Yes | 1 No |
Answer / kalyan
Run ur session in Verbose Mode and check out the Busy
Percentage in the Log. IF its more at the Reader thread
than ur Source Query is the Bottleneck.Tune your SQ.
If its Writer thread, then you check with your target . May
be you need to drop and recreate the Indexes on the target
table.
If its the Transformation thread , then check with your
mapping logic. Concentrate More on Aggregator part..
Fine tune your logic. Don't drag the fields which are not
used to all the transformations. try to use as less
transformations as possible.
Cache your lookups .Whenever possible use the persistent
lookup concept.
This should help guys..
| Is This Answer Correct ? | 17 Yes | 2 No |
Bottleneck in Informatica
Bottleneck in ETL Processing is the point by which the performance of the ETL Process is slowr.
When ETL Process is in progress first thing login to workflow monitor and observe performance statistic. I.e. observe processing rows per second. In SSIS and Datastage when you run the job you can see at every level how many rows per second is processed by the server.
Mostly bottleneck occurs at source qualifier during fetching data from source joiner aggregator Lookup Cache Building Session.
Removing bottleneck is performance tuning.
| Is This Answer Correct ? | 3 Yes | 0 No |
Answer / srinu
identification of bottelnecks
target:configuring session to write to flatfiletarget
source:add filter t/r after sq t/t to false show that no
data is processed past the filter t/r,if it time takes to
run new session remains same to the original session there
is source bottel necks
mapping:add filter t/f before each target and set filter
condition to false,similar to source
session:use the collect performance data to identify the
session bottel necks
read from desk,write to disk counters other than zero,there
is bottelnecks
| Is This Answer Correct ? | 3 Yes | 5 No |
Is it possible to create multiple domains on a single Informatica server?
What is aggregate cache in aggregator transformation?
any one can informatica metirial to me ple forwrd to my mail id; gbalireddy@gmail.com
my sourse is like id name sal--1 aa 1000 ,2 bb 2000, 3 cc 3000 4 dd 4000, 5 ee 6000 , 6 ff 7000 and so on but myrequirement is like this id name sal up_sal,1 aa 1000 null,2 bb 2000 1000, 3 cc 3000 2000 4 dd 4000 3000, 5 ee 5000 4000 , 6 ff 6000 50000 and so on so how can i get it plez reply for this as soon as possible , thanks in advanced
when we use lookup,if have 5 records,if i dont need 1st and 2nd records..what will be the procedure to achieve by using lookup?/
SRC1 -> EXP -> AGGR -> TGT SRC2 -> EXP -> Above is a maaping with two pipeline connected to the taret TGT. Design wise is this design is correct or not ?
how the data will be loaded while desiging the schema?which one first (for e.g.-dimesions and facts)
How did you handle performance issues If you have data coming in from multiple sources, just walk thru the process of loading it into the target
what is mean by flag 4 in update strategy t/r
how can we calculate fact table size?
What are batches?
How to display session logs based upon particular dates. If I want to display session logs for 1 week from a particular date how can I do it without using unix.
Teradata 
