Enough of the "DB2" thread, let's talk about Fragmented Tables

With the use of SANs for storing the database how effective will the use of fragmented tables be? I've got a few very large tables (16+ chunks) and was considering splitting it into 6-8 fragments, they all have serial columns and one has 7 indexes (or is it indeces). The last time I did anything real with them was splitting a table across 4 disks, it improved the performance no end. I'm not convinced I will get an improvement when using a SAN Regards Colin There are 10 types of people in the world, those that understand binary and those that don't sending to informix-list

"Colin Dawson" <cjd_1955@hotmail.com> wrote in message news:1122561967.47074756ec1baabe5e50e6d872905f6a@teranews... > > > > With the use of SANs for storing the database how effective will the use > of > fragmented tables be? I've got a few very large tables (16+ chunks) and > was > considering splitting it into 6-8 fragments, they all have serial columns > and one has 7 indexes (or is it indeces). The last time I did anything > real > with them was splitting a table across 4 disks, it improved the > performance > no end. I'm not convinced I will get an improvement when using a SAN > > > > Regards > > Colin Depends upon the SAN. I'm guessing here that you mean an EMC Symmetrix. In which case, in my humble-but-oft-criticised opinion, is that it's a waste of time for "normal" access, as the level of indeirection between the hypervolumes presented to the host and the actual physcial disk is so great, but still useful for parallel data query where Informix processes are taking advantage of the fragmentation.