One of the most important tasks of a BI Administrator is to ensure that regular backups are being performed.
This is normally done for disaster recovery purposes, but should be also done for other reasons of which the most important is forensic analysis.
An example of this would be when a deployment is about to be upgraded. Performing an extraordinary backup, that is specially labeled and stored, it will be possible to analyze the differences between the affected parts (settings, documents, connections, etc.) before and after the upgrade in case problems arise.
On the other hand, if the old content is overwritten, the investigation will be almost impossible, because the comparative analysis is lacking important elements.
Another important aspect of problem prevention is to use the Versioning features of Promotion Management (formerly called Life Cycle Management).
A scenario where one would be happy to have a Version management in plan, is when for unexpected changes happen at document or universe level.
For example: if accidental changes are done in a Universe and this is published back to the BI Platform, reports may stop working.
With accidental changes we mean both man-made and also those that can come by system failure (i.e. corrupted file or metadata)
In that case the BI Admin would simply have to roll-back the universe level and things would go back to normality.
The worst scenario would be when there is no previous copy of the universe, in that case there could be delays, stress and disruption to the reporting activities.
Last but not least is the need for tracking down changes that are done at any level of the BI deployment.
When support needs to investigate problems, it is important to understand what kind of changes have been done recently.
Because the extraordinary backups are closely tied to the change management process, I want to spend some words to explain what I would do if I was a BI Administrator in such sense. I would keep a log book where I would record any significant change to the BI deployment or to the elements that it is connected to (Databases, Servers etc.). Here too, if any problem shows up and needs investigation, who troubleshoots should be able to identify possible causes and discard those that are not pertinent to the problem.
In conclusion, the advice is to :
- Perform regular backups, i.e. :
– Every Day (or every N hours, if needed)
– [Quarterly, etc.]
- Keep a reasonable amount of backup copies
i.e.: last 7 days, last 4 weeks, last N months, all years
- Perform extraordinary backups immediately before any major change:
– Patch installation
– SP installation
– Version Upgrade
– Database upgrades and/or migrations
– Server migration
– any other major change one can think of…
- Label and keep the backup copy for a reasonable time
i.e.: until the next upgrade or change
- Use the Versioning feature
- Record all changes in a logbook:
– BI Settings (Server Parameters, Config. Files, etc. – if asked by Support, write down the incident number )
– Universe Structure
– Server settings
– Server upgrades (software, hardware, patches etc.)
– any other important change that may occur…
Hopefully the above indications will save some stress and time to those who at today have not considered this vital aspect in BI Administration.
These are my personal opinion based on experience, if you have suggestions or corrections you are welcome to express them in the comments section.
PS: For more information on Backup, Restore and Versioning, please read the Business Intelligence Platform Administrator Guide chapters 12,13,14 (for BI4.0 SP8) available at http://help.sap.com/boall_en