Why size of index area reduced after dump and load?

[balebob]

I noticed that after my first dump and load of the database with full index rebuild. The size of some index area is reduced to almost half of its size before d&l even though the data area for that index has never been archived. This is great but I would like to know why the size of the index area is reduced while its data area is at (almost) the same HWM as before d&l?
Thanks.

 

[TomBascom]

A few possibilities:

Did your index utilization increase? From, say, 40% to perhaps 80%?

Did you change anything else? Like maybe the block size, the rows per block or the blocks per cluster?

Are you sure that all of the indexes that were previously in that area are still in that area? And if they are did they all get rebuilt?

 

[balebob]

1. Yes, the index utilization increase over the year before d&l. In fact, I did a dbanalys before dump. Some of the table in the data area did have a very high scater factor (between 2 to 5). I ran another dbanalys after load and full index rebuild... and notice that the scater factor in the same area is now at 1.0. The total number of record before and after d&l are the same.

2. We use standard blocksize at 8192.

3. Yes, I am sure that all indexes were rebuilt properly. I did check my index rebuild log (I re-direct index rebuilt to the log file) and the database log... and I do not see any error. If there is an error, my user will knock on my door an hour after the db broght up ;-)...