Re: fast table scan

Depends what you want to do with the data, how much change occurs between scans etc, how busy the system is, how many spindles available, how many CPUs etc.

1.5 Mb off-disk - at a typical 64Kb per read request requires 24 I/O requests. Assuming disks capable of 10 millisec response per read request you can an idea of how unlikely you are to meet your requirement.

For suitably simplistic requests, with very little update taking place elsewhere (i.e. to avoid need to read rollback segments as well), you may have some success after the 1st pass with parallel query, several CPUs, lots of spindles, making the table a cached table, running on file system with a large filesystem buffer.

If you want to return 10,000 separate rows into a Pro*C array then that's another issue.

--

Jonathan Lewis
Yet another Oracle-related web site: www.jlcomp.demon.co.uk

rosmit_at_acxiom.com wrote in message <7f0atu$lq5$1_at_nnrp1.dejanews.com>...
>I am looking for the fastest possible way to scan a table. The table is
>small - 10,000 rows with a datafile of about 1.5MB. I want to run a query
>(or have ProC fetch) in .020 seconds or less.
>
Received on Fri Apr 16 1999 - 11:20:45 CDT