DSXchange

I would like to know the different ways which are generally used for performance tuning of Server jobs. Kindly reply me in this regard.

Welcome aboard. :D

The key performance indicator in ETL is elapsed time. If you can get the data in - or out, as the case may be - within the allowed time window, with a safety margin, then your ETL is performing satisfactorily.

The strongest mechanism for tuning that performance, therefore, is expectation management. And education that:

• correct is far more important than fast and that that is where you should concentrate your efforts
  
  rows/sec is an almost entirely useless metric

muruganr117 wrote:I would like to know the different ways which are generally used for performance tuning of Server jobs. Kindly reply me in this regard.

There are many thing you can do but it depends solely on your requirements. some of the points are as below.

1) Supplying a sorted input to Aggregator stage
2) Enabling Cache for the hash file
3) Avoid using joins in the source query if you can replace it with hash file and look-up
4) If possible divide the extraction and loading process in a job into 2 different jobs by landing the data in intermediate seq. file and loading the target from the file.

Manage expectations.