AM yesterday: Press release issued: “Two years of 100-percent uptime at 365 Main’s San Francisco facility”...
SAN FRANCISCO, July 24 /PRNewswire/—365 Main Inc., developer and operator of the world’s finest data centers, has provided online retailer RedEnvelope with two years of 100-percent uptime at 365 Main’s San Francisco facility.
PM yesterday: 365 Main datacenter power outage brings down major websites
It seems that the Web 2.0 datacenter 365 Main, in the heart of SOMA, has lost it’s power. Sites that are affected include Craigstlist, Technorati, Yelp, Netflix, AdBrite and all Six Apart properties, TypePad, LiveJournal and Vox.
Spooky. [via Metafilter]
Side note: For a bit, The Six Apart Twitter stream was one of the only places information was available on the issue. Will we see more companies launch Twitter streams for “just in case” scenarios?
AdamA
on 25 Jul 07Somebody forgot to knock on wood…
Rick
on 25 Jul 07Or maybe companies will host their servers in data centers with backup generators.
Chris Carter
on 25 Jul 07Or host their datacenters in states that don’t have a rampant hitory of power problems.
coughWashingtonStatecough http://www.datacenterknowledge.com/archives/quincy_wash-index.html
Sorry, I have to plug my own state every time talk of data centers comes up :)
Stewart
on 25 Jul 07Must be the name in it or something … Irish hosting company hosting365.ie went down for a few hours recently, apparently taking about 30%+ of irish websites with it (http://www.enn.ie/article/47482.html). They also ahve an external “support” blog, though the most up-to-date info at the time it was down came from community boards and mailing lists
Nick
on 25 Jul 07Word on the street is that it was caused by an employee:
http://valleywag.com/tech/breakdowns/a-drunk-employee-kills-all-of-the-websites-you-care-about-282021.php
Greg
on 25 Jul 07Making a claim like that encourages Murphy’s Law to spring into action.
Dr. Pete
on 25 Jul 07How can anyone claim to have 100% uptime? Were they absolutely sure their server was never down for a few milliseconds or inaccessible due to CPU consumption or an SQL index lock? Maybe they were punished for their hubris.
Vaughn
on 25 Jul 07“Will we see more companies launch Twitter streams for “just in case” scenarios?”
Well it’s not like Twitter is somehow magically immune to power outages. It’s just good luck they didn’t happen to use a SF colo for their service yesterday.
John Topley
on 25 Jul 07I heard they’re changing their name to 364 Main ;-)
Anonymous Coward
on 25 Jul 07“How can anyone claim to have 100% uptime? Were they absolutely sure their server was never down for a few milliseconds or inaccessible due to CPU consumption or an SQL index lock? Maybe they were punished for their hubris.”
The downtime resets the servers there, which means an angry mob of clients would be at the door waiting to restart / reconfigure their machines & services like yesterday. When they say 100% uptime, it’s pretty much fact.
Eric
on 25 Jul 07I thought that companies would be smart enough to have their “status” channels ( like status.sixapart.com ) on a completely different set of infrastructure for exactly this reason. Oh, and that means that the DNS has to be redundant too, just incase you forgot.
The fact that these sites went dark was bad design and a failure to plan.
Dr. Pete
on 25 Jul 07@AC – Having been on the receiving end of just such an angry mob, I understand the industry definition, but as soon as you say “100%” it sort of ignores the margin for error and daily fluctuations. I can accept calculations like 99.95% uptime based on actual, logged, server/connection outages, but as soon as you say 100% you imply that there’s no error bar and you’ve achieved perfection.
Warren
on 25 Jul 07Going back a few years here, but i remember a bbc article where they were talking about the great safety record of concord (never had a crash)
This may have been it – i cant remember exactly http://news.bbc.co.uk/1/hi/uk/848775.stm
Anyway a few hours later this happened http://news.bbc.co.uk/1/hi/world/europe/850903.stm
Spencer Lavery
on 25 Jul 07News creating news and all that.
Rowan
on 25 Jul 07agree with @Eric. Was very surprised that such large sites bit the dust unceremoniously.
Geographically/financially separate server(s) serving your ‘Woops we’ve had a problem page’ that comes into play when your main server farm falls over is the wiser idea. Comparatively cheap as chips.
I loved this quote from Technorati dude Dave Sifty ”...never mind that they always swear up and down that this kind of event can’t possibly happen, oh no, they have multiple redundant systems and they charge us up the wazoo to make sure that we’ll have business continuity, so of course, this isn’t really happening, oh yes”
Business Continuity on one infrastructure supplier ? Ya gotta be kidding me…how many offices have multiple internet feeds to their buildings for that very reason ???
Eric
on 25 Jul 07Was RedEnvelope down? The first article was all about RedEnvelope being 100% available for 2 years, not the other sites listed in the second article. The second article doesn’t list RedEnvelope as one of the sites that was down due to the power outage.
Ben Nevile
on 27 Jul 07I was in SOMA the day this happened. San Francisco as a whole had some weird power stuff going on.
Maggie
on 28 Jul 07“Hey, where would be the absolute worst place to have a data center?”
“Ummm… California… power issues…”
“Specifically?”
“The Bay area. Severe power issues. Both lack of and reliability of.”
“Cool. Know one in the area I can call for my hosting needs? ...”
I'm a jerk
on 01 Aug 07asdgsaasgsgf
This discussion is closed.