Re: Oracle process timeslice,setting priority

I have several largetables, all being a single month of data primarily consisting of many 'strings', and lots of bfiles/blobs. Each month of information has, for the past two days, taken: 1. As a single query - 14 hours.
2. As I have written it with the cursor/loop - just under 21 hours.

There are 19 months of data. One of the reasons I have opted for the cursor approach is that within the loop I do a commit. Thus, if the job has to be stopped, which has been the case of late because of the timeslice problems, I haven't wasted the processing time prior to the drop & can extract out the metadata I've collected thus far. One of the other reasons I've opted for the cursor approach is that the block of code is a lot more complex than I have written in my original posting. For the entire collection of metadata I am extracting there is a bit more going on that can't really be represented as a single query.

David

Galen Boyer <galenboyer_at_hotpop.com> wrote in message news:<u7kedsy8h.fsf_at_standardandpoors.com>...
> The issue is that the code is poorly written.
>
> Why don't you rewrite this to do the update in a single query?
Received on Sat Dec 14 2002 - 03:30:26 CST