Jaiku Is Down: A Symptom of a Big Web 2.0 Disease
by
on February 18, 2008,
Still half out of it recovering from the flu, I caught a few blips crossing my Twitter stream about Jaiku being down. I yawned, assumed some server farm somewhere tanked, and haphazardly clicked a link to check it out.
Yep, Jaiku is down. Only, it's not. The site is up, which means something is working somewhere, but the reason is “two disks kicked the bucket.”
Occasionally I have to consult with people still doing web app development to see how far back in the dark ages my knowledge of best practices now resides, but “two disks kicked the bucket” seemed like a poor excuse for an outage. My consultation confirmed that things hadn't changed that much.
I've frequently railed about the problems with many of the Web 2.0 apps that are out there from a technical perspective, but Jaiku's outage is a glaringly obvious example. If you don't build an app to scale properly at the outset, you need to bag your code and do it the right way. This tendency to slap a Band-Aid® on each little problem and cobble something together isn't an approach that can work in perpetuity. At some point, your app has to just work, and keeping the “beta” tag on your logo isn't a free pass.
I'm sure there are those who will compare last week's Amazon S3 outage to this, but the two aren't even remotely similar. A data center outage is not only a rarity, but usually due to some unforeseen circumstance that is not only rectified, but is then followed up on to make sure that same circumstance won't occur again (like Amazon is doing with the issue of authenticated calls that brought S3 down). You can't consider hardware failures unforeseen; hard drives fail all the time. If you built your app to distribute the data held in memory across multiple servers, with, say, an archive process to an actual database, if one server fails, the site doesn't go down.
Why don't more folks developing apps start out thinking about what might take their app down? The little-guy shtick only goes so far; at some point, you still need to view yourself as a professional. An app that can't handle a drive failure screams amateur.
If you enjoyed this post, make sure you subscribe to profy RSS feed!
Svetlana Gladkova
Peet McKimmie 
65bus 
Irel Johnson
Google’s orkut still has the “beta” tag on it still after four years. They need to get their shit together on that one as well.
Orkut has largely been a flop in the US. It caught on in Brazil, but had serious issues with sexual predators targeting kids. However, for all of its issues, I don’t remember stories about the service being down, because Google knows how to keep things running.
I don’t use Jaiku so can’t say anything about it but in general, I think this is an interesting question. For me personally, it all depends on the context. How much do I like the application? What happens when it’s down? How do they interact with their customers?
I find Twitter’s constant outages a bit irritating. But I love Twitter, and I really like how Twitter as a company talks to me. So it doesn’t bother me. Even Dreamhost bumbling along as they do - I take it in my stride because I really enjoy interacting with them.
But how long will they be able to run on my good will? I hope these guys DO get it together.
Dude, I’ve gotten so used to seeing “beta” on logos now that it doesn’t even register. I’m more likely to notice an alpha or, horror of horrors in this Web 2.0 world, a full release?
Isabella, you raise an interesting point, but one that I think it integral to a lot of the tolerance of this sorts of outages: user good will. In the stage many of these companies are in, it’s a free service so you tend to putter along thinking “Hey, Twitter is cool, and it’s free, and so what if they are down a lot?” How far does that good will extend if you were to have to start paying to subscribe to Twitter?
Referencing Dreamhost goes along with that same mentality, too, I think. They are cheap. Dirt cheap, so the outages still fall in line with “you get what you pay for.” I have a fabulous host for my own sites I’ve had a relationship with for over 5 years now, and in the entire time I’ve been with them, I haven’t had as many outages as friends I know with Dreamhost had last year. So I guess my question now is, what is the price point at which good will covers poor or fair service instead of excellent?
It’s about way more than price. I really like how Twitter and Dreamhost treat me.
The whole pricing thing is so interesting. For example, I’m organizing two meetups here in Vancouver. Meetup.com is a clumsy application, kinda old-fashioned, and I’m not particularly taken with their customer service. But I love what it lets me do, and how it brings people together. So I dish out the $15 a month, which of course is a ridiculous price when you think of it.
Now would I be willing to pay $15 a month for Twitter? Honestly, I think not. Even though I really like Twitter. Isn’t that weird …