Scaling Fast | Hacker News Reader

Scaling Fast

lukasbiewald.com

107 pointsluccastera16 comments

Been there, done that. :-) And for those about to downmod me, it was digg in Feb. 2005 and Paris Hilton's cell phone got hacked, and we were the no.1 and no. 3 results on yahoo search for "Paris Hilton Cell Phone" out of the blue.

It sounds like they did a better job than I did.

Is your experience shared in text form somewhere? If yes, please post; better yet submit it. If no, please post, and better yet, submit it :)

Nice site. I did spot an XSS bug:

http://facestat.com/faces?message=%3Cscript%3Ealert(%22test%22)%3C/script%3E

How much money did the ads bring in?

This really scares me, because I have a site with the exact same beginning architecture on the exact same hosting provider..

I guess I had best get to work.

Please don't get to work scaling your architecture yet. Work on making your app awesome and wanted by lots of people.

Trust me, I've been bitten by this problem myself. It's so easy to get really worried about the number of hits you're going to take. It's also so easy to get really excited about the scaling challenges.

But, quite frankly, the chances of you getting featured on yahoo.com are miniscule! First focus on getting there by creating something cool. Then focus on scaling.

That's what I've been doing so far. I'm trying not to fall into the trap of premature-scalingitis. But the original article points out that the need for scaling up does not always come at a time of one's choosing. And that's kind of scary.

And then as if the gods of scaling were taunting me, right as I posted this, my front-end web server ran out of memory and started thrashing. I had to hard-reboot it to recover.

I think the key here is that their architecture allowed for a bit of scaling. How many companies don't have that?

You were given a chance and you capitalized on it because you could. I wonder how many other companies failed to capitalize on their chance and went bust.

Really I'm not sure what you mean - only the most naive web architectures wouldn't allow for some type of scalability. The way they did it was with the old "Throw More Machines At It " method. I'm not sure this ever really shows a mo better web arch - even spaghetti PHP can do that pretty easily.

Wow Lucas, you put our scaling story to shame.

When are you guys coming by and visiting dude?

I can't believe your comment got downmodded. I'm biased though, because I read your story.

Thanks for looking out, lol.

Does anyone think it is worth it to set up your site/service on EC3 with something like Scalr, so your site automatically expands to handle the load?

It would take time away from adding features and delay your launch, but it may help you handle huge bursts like this. But then again, it could slow down your development enough that your competitors get the huge burst, and by the time you're ready your niche is already passe.

With the ridiculously quick news cycles we have today you may only get one chance - so is the ability to capitalize on it worth the cost?

EC2

two thumbs up to slice host :D

Yeah, that's pretty much the perfect problem. Your servers go down since you're getting too many hits.

Reply on news.ycombinator.com