Ensight – Jeremy Wright

Twitter is currently down. Has been for awhile. Scheduled maintenance, so this isn’t a big deal. Except that, as Dave Winer recently, scheduled downtime of 6+ hours on a service people rely on for communication isn’t acceptable. Stil, like Dave, there are certain situations in which you need to pause writes to the database, so if Twitter is doing a major architectural update… Well, they won’t have much of a choice here.

So that’s the wrong way to do downtime: long outage for something people rely on to communicate. Hell, if Twitter was up right now I’d normally be twittering this!

But, that said, major, major kudos to the Twitter team for their hourly updates on what’s extending the downtime, visible here at their blog:

Update (12/17 12a): We ran into a minor glitch and will be taking 1 more hour of down-time.  Thank you for your patience!

Update (12/17 3a): We’re slowly bringing services back up.  You’ll probably see some slowness for a while.  We hope to be back up to speed before your finish your bagel and coffee, NYC.

Update (12/17 5a): We’re still working on fixing some issues causing massive slowness site-wide. We’re on it.

Update (12/17 6a): We’re waiting on a fix to our network switch which we’ve determined is causing the slowness we’re seeing.  More soon. 

The updates totally calm folk own. Major lesson for all tech geeks and folk who keep sites up: Yep, when things are down you have to work your ass off to bring’em up. But dropping a quick hourly note out to the folk who use your service will keep them pretty calm and forgiving. Ignorance isn’t bliss