<I've had a quick look at the ADS indexing web page and it looks as though they follow a different approach to Tim so its not a like for like comparison.>

I guess you are right, 2 different products.

<I would have expected a bigger drop than that moving from unicode to ansi so the first thing is to check that they have all been switched to ansi. After that about all I can think of to suggest is to review what each index is required for and if its still needed.>


This is a very good aproach to recoding, i found a lot of useless indexes right now in the old database, so it is a good time to improve things.

Thanks for your ideas.
best regards.

Jorge,

<< The resulting data is 2,8 times bigger. >>

It's the indexes - ElevateDB only uses duplicate-key compression, whereas ADS probably uses trailing-byte/suffix compression also (or a combination).  Our other product, DBISAM, uses these kinds of compression, and what we found was that the compression added more overhead than simply reading from disk, so we ditched this in favor of the simpler, and faster, duplicate-key compression.  However, what this means is if you use string keys for primary/unique keys, the resultant indexes will be quite a bit bigger.

Tim Young
Elevate Software
www.elevatesoft.com