Indexing is taking too much of time.

[Former Member]

Hi,

Processes chain is getting slow from midnight(12) to 3 am.Other than this timings it is working good.(it runs after every 20 m)

According to my analysis indexing is taking to much of time for that period.

Please suggest possible way to slow it down.

[Former Member]

Hi Ompreet,

As you said it runs every 20 minutes and it runs fine all the time except for the duration 12am to 3am, please check if there are other heavy background jobs running or other process chains running which are consuming the work processors.

If it is getting delayed only during a certain period, then this might be the reason.

Also consider the time gap as Suman stated.

Regards,

Mohammed.

[Former Member]

Hi,

Check if many other processes are also running at the same time.

Also you can check the indexing log for the different times for comparison of new Vs old log.

This might give you an idea to go ahead.

Rest as all have told, indexing will depends on the amount of data volume being involved.

Br,

Arpit

[MGrob]

Hi

Droping and recreating index will get slower the more data volume you have in your cube. What we did is building index only new on the weekend in a maintenance chain. During the week we just load data and keep the index as it is.

These days you can make use of the housekeeping list which recreates all indexes in a background job (also a weekend thing) see my doc http://scn.sap.com/docs/DOC-46240

hope it clarifies

Martin

[Former Member]

So it will make processes fast right?

[former_member182470]

That's what I told you to remove those steps from the chain and schedule them in off-peak time in other period like weekly or 10 days or  fortnightly.

This obviously make processes fast.

[former_member182470]

The time for creating indexes on a Cube will be dependent on the volume of Cube data. How big is your volume?

How many delta records you are getting into your Cube on every 20 Minutes basis? If your chain is running every 20 Mins, then it is difficult drop all the indices and rebuild again within just 20 Mins of time before the next run. The time gap is too less.

If your delta size is not so big, then i suggest you not to remove that step in the chain. You have to remove drop indices step also.

Fast indexing can also depend on the database also.

Regards,

Suman