I have a v4 table (on our win2k file server) with about 197k records
that are 77 bytes wide (according to the tdr utility). The table uses
4k blocks and has a file size around 133mb. The problem is that ".."
qbe queries do not appear to use the available secondary indexes. For
example, a ".." search in the field with last name takes 12 to 15
seconds on my workstation; searching the field with first name takes
the same time--but there's an index on last name and none on first
name. A query on last name *without* ".." is < 1 second. On the
machines used by actual users, time is more like 2:30 for ".." and just
a few seconds for no ".."--their machines are on the 10mb network and
mine is on the 100mb network.
No specs I know about v4 tables suggest I'm anywhere near a tipping
point. Still, this slowness just began (we *have* had a few earlier
episodes--some cured with reindexing, some from network problems), and
we can't find anything wrong with the table, the client machines or the
network. So I suspect I don't know *something.* The thread in this
group titled "DOS - Determining maximum allowable size" seems to hint
that secondary indexes have problems after a table is more than half
its theoretical max. size.
So maybe someone would please tell me more about this?
Oh, there's another fact that's really interesting: I don't write
pdox4.5 PAL anymore--I only maintain. I have a BDE app (using BDE 5.x)
that queries the same table with the sql equiv. of ".."--the "like"
operator. On my machine, results return in sub-1 second time. That
suggests that *someone* can use the secondary indexes successfully. ;-)
Thanks for your help.