Re: I/O performance of SQL SEQUENCE objects

Hi Charles,
@GUID: it was only unique per processor, at a multiprocessor box, we've got collisions.
@own function: I wouldn't use a dtaara and recommend one table for multiple "sequence" the diffrence between the sequence and a function of your own is:
- you don't need the embedded SQL layer
- the level of blocking is better adjustable (you could get 20 Ids with one call and use it inline)
- for multiple processes, needing only few numbers each, the get function could be dataq based (one resident function to deliver numbers for n processes
The fastest solution I did mention in my previous post.

Dieter

<Charles>
Lol...I recall those conversations. Shouldn't be an issue, as a GUID would
only have to be unique within the machine.

Replacing the use of a sequence object with our own data area is an
option. Though the question is, can a straight RPG program perform the
operation faster than the DB can with a SEQUENCE?

I'd like to think IBM's optimized the SEQUENCE object and that RPG can't do
it any faster; but I suspect that may not be the case. I've just never
seen any benchmarks.
</Charles>