A few folks are commenting about one of the problems of syndication, specifically the fact that at the top of every hour or so, they get hit by a barrage of RSS requests.
Honestly, i look at it from a slightly different point of view.
First off, what are you trying to accomplish by using Syndication?
This is a legitimate question that i'm damn sure a good many folks just don't understand. Syndication (be it RSS/Atom/CDF/ or whatever) serves two roles, one is notification the other is distribution.
By Notification, i mean informing people who ask that "Yes, i've got something new." For the record, this is how i tend to use Syndication. i run a little reader plug in Firefox called Sage (the update to RSS Reader) which allows me to do a general request for new content. It doesn't automatically poll, which was annoying at first, but i've learned is actually preferable. This way i'm not distracted by notifications of content, but can read them at my leisure.
The other role is Distribution. Using most Syndication mechanisms, it's possible to provide most if not all of your content. This can be helpful if you're providing content to remote sites or aggregation portals, or if you're just being a nice guy.
i'd also not that these roles are not exclusive, and in fact, are often mixed. The question then becomes "What do i want my audience to do?"
Personally, i want my audience to come here to my blog so that they can read and participate in whatever stupid drivel i post. To that end, i generally make the abbreviated RSS feed available. This is a feed i've created that has the least amount of information i feel i need to provide for folks (leading excerpts of the last ten posts, no HTML formatting) . This is the feed i provide automated crawlers and other bots.
i also provide a fully formatted distribution because folks asked for it. i provide that mostly because i don't expect those folks to come to the page, but of course, they will anyway.
If i started seeing the kind of load averages that would make me wonder about the value of offering RSS to the public, i'd solve the problem just like i'd solve the problem for any other web service, i'd spawn it off onto it's own server, just like Mail and HTTP are on separate servers. If it backlogs and times out, big deal, my main server is still there and folks would hit that instead, or they'd simply pick it up on the next refresh. The savvy folks would also be smart and pick some random, non :00 time to go fetch data and normalize my load distribution anyway.
i dunno, i just don't see what the big deal is. Folks, this is protocolled content, just like the stuff you're normally producing. If you're having scaling problems, don't blame the protocol. (Because, obviously that HTTP protocol is crap and simply can't scale.)
