Skip to main content.
home | support | download

Back to List Archive

Re: Append to Index File??

From: Jordan Hayes <jmhayes(at)not-real.speakeasy.net>
Date: Tue Sep 06 2005 - 14:28:40 GMT
> Doing a "merge" for hundreds of web sites is tedious.

The common idiom for this is to break your updates down into smaller
groups, and have your searches check multiple indexes.  For instance, I
run a mailing list with about 600 messages per week.  Every 4 minutes I
rebuild the index for "this week's" messages; once per week I reindex
"this year except for this week" and once per year I reindex "all time
except this year" ... my searches are on three indexes, I spend very
little time rebuilding, and I haven't touched the cron jobs in almost
two years :-)

In the fall, the "tihs year except this week" begins to take longer, but
I think the worst it gets is about 15 minutes once per week.  Not too
bad.

But: I think you have a different problem, since you need to delete
pages too, right?  You might want to do these macro merges with a
different metric, trying to optimize for not rebuilding pages that
aren't changing much.  In the long run, this might not scale for you.
Swish-e isn't a serious competitor to Google ...

/jordan
Received on Tue Sep 6 07:28:45 2005