[dokuwiki] Re: The database question: a new spin
- From: "Joe Lapp" <joe.lapp@xxxxxxxxx>
- To: dokuwiki@xxxxxxxxxxxxx
- Date: Wed, 14 Sep 2005 08:09:11 -0500 (CDT)
>From: Seun Osewa <seun.osewa@xxxxxxxxx>
> What features exactly, would benefit from the database?
The main benefit is that you improve server availability, increasing the load
that a given server can handle before connections start being denied and
browsers start timing out waiting on connections.
If the database 'cache' has been built, you will probably also get some, mostly
negligible, increase in page load times.
However, the improvement to RSS feed generation would be non-negligible on both
accounts. On busy sites, the RSS feed could be new with each new request for
the feed, making it hard to benefit simply by caching the previous feed. But
the feed has to be generated with bits of information collected from many
pages. Right now the feed loads all of those pages from disk, though it only
parses pages not previously cached. We could store these bits in a database,
reducing disk access requirements, and if the database caches recently
retrieved entries, possibly virtually eliminating disk access.
Link color determination is another area that would benefit. I doubt anyone
would notice a change in page load times, but this disk-intensive activity
could be reduced to a nearly diskless activity, and that probably translates to
a huge boost in availability.
And full-text searches could also benefit, depending on how they're
implemented. I don't have any experience implementing full-text searches on a
database and so can't say whether they could actually be made faster or less
resource consumptive -- but I suspect that a database would help a lot.
Again the main issue is peak-load performance or server capacity or
availability -- what ever the word is.
~joe
--
DokuWiki mailing list - more info at
http://wiki.splitbrain.org/wiki:mailinglist
Other related posts:
