Comments on: Hey Twitter, Maybe It’s Better Not To Share Absolutely Everything

By: tandemrandom.com » Twitter And FriendFeed Battle For Downtime. Scoble’s Head Explodes.

Sat, 09 May 2009 08:54:06 +0000

[...] completely lost contact with its data center, co-founder Bret Taylor tells me (hey, Twitter once lost a database, so this is nothing new). They’re investigating the problem right now. But where are we [...]

By: Twitter And FriendFeed Battle For Downtime. Scoble’s Head Explodes.

Twitter And FriendFeed Battle For Downtime. Scoble’s Head Explodes. — Sat, 09 May 2009 00:03:44 +0000

By: blog-thing : Twitter Cans UK SMS

blog-thing : Twitter Cans UK SMS — Thu, 14 Aug 2008 11:47:03 +0000

[...] questions have been asked about these performance and downtime issues. Do you really want to tell the entire world that you’ve lost a database? Twitter’s recent issues with replies were also examined and the suggestion is that [...]

By: Infotron » Blog Archive » Apparently You People Don’t Really Care About Twitter Downtime

Wed, 09 Jul 2008 01:36:39 +0000

[...] the site more than once in thirty days) has held steady at around 55% since March. Twitter has seen near-constant downtime and disappearing features in that time, but nobody seems to mind. At least, not enough to try out [...]

By: Apparently You People Don’t Really Care About Twitter Downtime

Apparently You People Don’t Really Care About Twitter Downtime — Tue, 08 Jul 2008 23:22:26 +0000

By: web

web — Sat, 07 Jun 2008 13:22:17 +0000

Good points, but make it more generally

By: The Twitterless Week Experiment « TechTicker

The Twitterless Week Experiment « TechTicker — Sat, 07 Jun 2008 10:45:46 +0000

[...] night between database failures. Everything seems back to normal now. Perhaps somebody returned that lost database to [...]

By: CVOS SEO

CVOS SEO — Sat, 07 Jun 2008 06:35:18 +0000

Like a figure from Greek mythology, every arrow twitter takes from Arrigontington’s bow just makes them stronger.

By: JC John SESE Cuneta

JC John SESE Cuneta — Sat, 07 Jun 2008 06:26:11 +0000

ROFLOL!!!!

I think I found their database floating in the Pacific Ocean… recent photo from Google Earth, with sharks circling it…

—
Yeah, they are giving too much already, or at least re-word it. hehe, but it was funny, maybe they intend to make us laugh

By: Trace

Trace — Sat, 07 Jun 2008 02:08:28 +0000

@Ivan: I guess you missed the part where this has been going on for years.

By: links for 2008-06-06 « 個人的な雑記

links for 2008-06-06 « 個人的な雑記 — Fri, 06 Jun 2008 22:46:03 +0000

[...] Hey Twitter, Maybe It’s Better Not To Share Absolutely Everything (tags: twitter) [...]

By: Twitter Suffers Minor Period Of Uptime

Twitter Suffers Minor Period Of Uptime — Fri, 06 Jun 2008 22:29:39 +0000

[...] night between database failures. Everything seems back to normal now. Perhaps somebody returned that lost database to [...]

By: Ivan Pope

Ivan Pope — Fri, 06 Jun 2008 22:29:24 +0000

Do you remember when Google was young and growing at a huge rate and it used to go down all the time because they only had three servers and they couldn’t work out how to scale it and keep it up and work out the architecture all at the same time? No? Me neither. I don’t think this is really necessary.

By: GeekMommy

GeekMommy — Fri, 06 Jun 2008 21:04:42 +0000

This is what you get when you get them to understand that they need better communication and more transparency, BUT they don’t get someone to be ‘that guy’ who isn’t a geek.

Hint Twitter: “We are presently experiencing an issue with one of the Database Servers – we are busy getting it fixed and should be fully operational again shortly. Our estimated recovery time is presently: XX mins/hrs.”

By: Luke Krogh

Luke Krogh — Fri, 06 Jun 2008 18:32:08 +0000

Hahaha, Thanks for the laugh!

I completely agree with Arington here. Not only is this unprofessional, but it’s also a security risk. They are getting way too technical with their information given to the public.

By: anon

anon — Fri, 06 Jun 2008 17:55:38 +0000

If what was posted in the previous article is true, they should really start clustering their DBs. Replication is not a good solution given their size.

By: Does Social Media Site Reliability Matter? - Covering All That's Social On the Web

Fri, 06 Jun 2008 16:35:46 +0000

[...] frequently reoccurs is that “Twitter is Down.” Just last night Mike Arrington posted an error message from Twitter which said “We just lost a database about 5 minutes ago and this has severely [...]

By: jro

jro — Fri, 06 Jun 2008 16:23:26 +0000

Lost a database, scalability/uptime, Scoble’s umpteen-billion followers, etc. Non-trivial engineering tasks to solve here.

Not to overstate, but the barrier to entry in solving these issues using things like Google App Engine seems rather low. I’m sure there are things I don’t know about all the features here, but the cost involved with GAE lowers the risk tremendously.

Twitter has a good problem in that their problems are due to high popularity, but it sure seems the underlying service could be substantially commoditized.

By: Peter Urban

Peter Urban — Fri, 06 Jun 2008 15:47:52 +0000

That’s a typical case of geek talk vs. the rest of the world.

Geek talk: ‘We’ve lost a database’ = on of our countless machines is down and it’ll be back up in a few minutes.

Rest of the world: ‘We’ve lost a database’ = The database is gone forever and we’ve lost all your data and now you’re f***ed – there is no need for you to ever come back again.

Sorry, I am all for open communication directly from the source but if you let the geeks talk you need to give them a quick tutorial how the rest of the world translats what they say.

Maybe a micro marketing department is in order… (I know, I know, that’s so not web 2.0)

By: homaid

homaid — Fri, 06 Jun 2008 15:38:21 +0000

friendfeed is so much fun than twitter, it help us build a better social-networks experience following friends on stumbleupon, digg, delicious & twitter, while building your network on friendfeed, so cool.

By: stationstops

stationstops — Fri, 06 Jun 2008 15:31:32 +0000

The fundamental idea of redundancy is that a single node failure *does not* affect site performance.

By: thecolor

thecolor — Fri, 06 Jun 2008 15:21:39 +0000

Perhaps that explains why I keep seeing old users I followed (and no longer) popping up in my list. I keep having to un-follow them, which (if related) makes me think the database really needs a map.

By: sat_

sat_ — Fri, 06 Jun 2008 15:12:07 +0000

@mat,
this joke is on bash.org since always

by the way, they really should consider MMM replication
http://code.goo...-master-master/
provided that they are using mysql.

By: Gebadia Smith

Gebadia Smith — Fri, 06 Jun 2008 15:07:53 +0000

I find it amusing how Mike is so willing to point out perceived flaws in other companies yet he isn’t willing to open his new tool bar up for comments.

Gary Roberts AKA the Chuck Norris of Hockey wouldn’t be afraid of a few comments.

By: EH

EH — Fri, 06 Jun 2008 14:36:03 +0000

Technical people would not have a problem with Twitter’s phrasing here. It’s only business types who need to have their hands held and reassured in soft tones, “it’s not the end of the world.”

Next they’ll say one of their servers “took a crap” and TechCrunch will post an article wondering whether Twitter puts toilets in their datacenters.