We recently completed data conversion for next release in our production environment. Already there are lots of good blog posted in SDN about data conversion. In this blog I would like to give few tips from basis side.
Running a data conversion in a production environment is a challenge for basis team. It becomes more crucial if you have release roll-outs where you need to manage existing load + data conversion on production server.
From basis side conversion life cycle at very high level has three stages:
Basis needs to keep close eyes on various systems during conversion run to analyze the following:
1. Number of background processes: How many background processes required
to complete the conversion during planed time frame.
2. Number of upgrade processes : How many UPD and UPD1 upgrade process is
required.
3. Workflows : If any unwanted workflow can be switch off.
4. Lock table entries: Verify lock table setting is sufficient for conversion and
regular user load on system.
5. Physical Memory/CPU: Additional memory and CPU requirement.
6. Buffer/Memory setting: Extended memory and buffer settings.
7. Database tuning: if temporary index can improve the conversion performance.
Once above points are finalized basis needs to create optimal conversion schedule in production environment considering the following:
Month end:
During month end we need to lock the users and transactions which will impact the conversion jobs, make sure conversion and month-end should not run in parallel.
Payroll:
Payroll is very important for any company, running of payroll job longer than expected schedule due to conversion will cause major problems.
Regular user load:
Increase in user response due to conversion load will cause unnecessary tickets on basis team. To avoid this create operation mode adjusting conversion process in night and day mode.
Conversion:
Go-live is dependent on conversion completion. Make sure conversion schedule should be meet in any situation.