Message: 1 From: kband@www.llamacom.com Subject: Re: [Wikipedia-l] Everything in the last digest To: wikipedia-l@nupedia.com Date: Sun, 3 Mar 2002 21:54:40 -0600 (CST) Reply-To: wikipedia-l@nupedia.com
Why would we want to distribute hosting? And for
that matter, what's
wrong with advertising, provided it's done as
subtly as possible, and
not from various merchants of death ;-) (which I
still think is for
Bomis to decide -- I may feel very proprietary
about a lot of stuff
here, but I still remember I'm on somebody else's
playground. Jimbo et
al. have been very cool about trying to get
input, but business
decisions should belong to them)
It's a matter of scaling. Hosting is currently the only bottleneck in the wikipedia process, other than possible problems in organization of material. If the popularity of Wikipedia doubles, the hosting needs to double--with distributed hosting, that would happen automatically.
-tc
--__--__--
I understand that, Cunc, but wouldn't Bomis having larger, perhaps more redundant (and mirrored) servers work as well? I've never dealt with wiki technology, but my last tech job was with a firm that did online transaction management. We had three active servers up at all times (web, SQL, and mail), and were always running checks so that we could add more servers as soon as it was necessary. It's very expensive, but isn't that where the advertising comes in? I know remote networking is viable, but is it practical? It just doesn't seem to be a sensible solution long-term. Can you imagine having to deal with migrations, etc., every time someone decides to back out? Ugh.
---JHK
__________________________________________________ Do You Yahoo!? Yahoo! Sports - sign up for Fantasy Baseball http://sports.yahoo.com
Julie Kemp wrote:
I understand that, Cunc, but wouldn't Bomis having larger, perhaps more redundant (and mirrored) servers work as well?
Yes, and it's worth noting that we are a long long way from this really mattering. Traffic would have to increase by a factor of 10 before I would really start to care at all. I'm happy to throw another couple of servers at the system, if it is justified by traffic levels.
It's very expensive, but isn't that where the advertising comes in?
Well, we're running free software on commodity hardware. I'm happy to supply whatever we need for the visible future. So no money will be necessary from any sources other than me for a long time. It's not _that_ expensive.
I know remote networking is viable, but is it practical? It just doesn't seem to be a sensible solution long-term. Can you imagine having to deal with migrations, etc., every time someone decides to back out? Ugh.
I think it's very impractical and "pie in the sky" from a technical point of view. There's a lot of neat ideas out there -- freenet, etc., but we aren't really a *technical* project, we're an _encyclopedia_ project. So I agree with you completely.
Our great virtue, to date, is that we stick to what we know. Inventing astounding new distributed hosting solutions seems too far off our central mission. We should, of course, utilize the best available stable technology at any given point in time.
wikipedia-l@lists.wikimedia.org