Oh Noz! Mah Twittah Iz Daon
From is twitter dying?:
It seems that everyday now there is a problem with Twitter. Not too long ago I heard John C Dvorak say that if the public jumps on twitter it will go down. Is that what is happening? Today their main database, db006, went down. I always know something is wrong when Twhirl wont let me get tweets even if I haven’t gone over the limit.
It seems from the comments that people are pretty understanding when it comes to twitter downtime. I tend to agree with them, to a certain point. Free services such as Wordpress.com have brought expectations up on free services. Although, the downtime can be quite frustrating.
Why can’t they just get some decent hardware and produce some better uptime? I’m sure it wouldn’t be hard to raise some money.
Lets take a closer look at this:
“Not too long ago I heard John C Dvorak say that if the public jumps on twitter it will go down. Is that what is happening?”
Fearmonger.
“It seems from the comments that people are pretty understanding when it comes to twitter downtime.”
Look again, people are acting like children. Everything is normal.
“Free services such as Wordpress.com have brought expectations up on free services.”
11,000 nails per hour. That’s an old article. How many hits per hour you think it gets now? My money is on more. You think wordpress gets that kinda traffic? Doubtful. If for no other reason than the posts are usually longer than a tweet so people take longer to read them.
“Why can’t they just get some decent hardware and produce some better uptime?”
You are confusing the issue. They are in the middle of an almost complete architectural redesign. Also, can you imagine the size of the db? I can’t. You ever do a backup restore of a db larger than ~100mb? It takes a while.
I don’t mean to pick on you dude. Truth is, I picked your post because it was the first in the link list on man down. Please don’t take all this the wrong way. I think most of the panic we see about this outage is due to twitter’s new policy of trying to be more transparent about service problems. And for the record, I doubt twitter is dying. But what do I know.
on May 27th, 2008 at 8:35 am
I know twitter isn’t dying. I was bringing it up because they had to do unscheduled maintenance on their main database yesterday due to overuse. This reminded me of when I heard John Dvorak make that comment about the general public. And as for Wordpress, I’m sure if they got that kind of traffic Matt would jump on it pretty quick.
on May 28th, 2008 at 3:19 am
Thanks for the critique though
on May 29th, 2008 at 10:56 pm
I’m sorry if it seemed like I was jumping on you dude. Truth is, I thought your article was well written and because of that it would allow me to address the issue. Your post outlined very nicely the general “Twitter is Dead!!!!” paranoia that was running wild through the blogosphere. So thanks for allowing me a great place to jump into the conversation.
On a side note, I like your blog so I’ll be adding it to my blogroll and I’d like to encourage anyone reading this to check it out.
on May 30th, 2008 at 7:30 am
It’s cool and no problem, I’m glad you contributed to it like you did. I was surprised to see all the traffic that I got on that post. I guess the link from ‘man down’ helped a lot.
Thanks for the link. I just redesigned the blog not too long ago and had originally intended to hold off on the new look until i go with a new domain in about a month. When I do finally everything squared away with that I’ll be sure to send you the new link.
on May 30th, 2008 at 9:02 am
Just an update on the actual argument here — Matt was on the Wordpress Weekly podcast this week and he said they serve more than half-a-billion requests per day. That is over 20 million requests per hour.
on May 30th, 2008 at 3:53 pm
Thanks for the update Josh. From my original post “11,000 nails per hour. That’s an old article. How many hits per hour you think it gets now? My money is on more. You think wordpress gets that kinda traffic? Doubtful.” Looks like I guessed wrong.
As I said though, that was an old article and I honestly don’t know how many db hits an hour twitter is taking now.
If its a comparable number to wordpress then we have to ask the question “how is it different?”. Possible answers: hosting, number of employs (I know twitter has 16 but I’m unsure about Automattic), architecture and backend (ruby on rails vs. php). We know for certain that the dev community behind wordpress is larger and that certainly makes a difference. Perhaps it is the key difference.
Wordpress is open source and like any open source project with a large following they reap the benefits, you basically have a whole army of geeks testing, tweaking, and modding your code for free. This makes me look at gillmor-gang-050808 a bit differently. They were discussing nationalizing twitter and while that idea is silly, a distributed open source model is not. If twitter had the power of the OSS movement behind it, combined with its huge following (which would grow exponentially once JohnQ could run his own twitter install that had two way aggregation with 1) the central twitter server and 2) customizable networks of twitter servers) there would be no stopping it. The users would be much happier because their networks could ping each other even if twitter central was down.
I know some people will say I’ve gone way out there now and gotten too far away from what twitter actually is. But have I? Perhaps I have, it is almost 4am afterall.
Thoughts anyone?
on May 30th, 2008 at 9:35 pm
I had forgotten that twitter opensourced their messaging queue Starling.
I did a little googling for decentralized twitter and the very first result was very interesting, particularly the comment from jstanforth:
Jabber. Groovy.
And even better Blaine Cook has long endorsed the idea:
And that’s just from the first result in Google. 94,200 to go.
So… I’ll add some quotes and narrow it down to 5,850 I’ll do some more looking, but it seems to me that with this much out there we aren’t far away from seeing this implemented. I hope so anyway.