365 Main deserves a great deal of credit for publicly posting detailed information about the cause of Tuesday’s power failure (mirror) and linking to it from their home page. That is virtually unheard of in this industry. As much as I like to tout my own company, we don’t even do that — you won’t find anything about P1DH’s Atlanta power failure some months ago on dedicatedhosting.com and details of the more recent problem at ServerBeach’s Dulles facility were buried in their forums.
I’m calling us out on this, as a customer* and employee! And you should call out whomever you host with too. There is a natural tendency want to bury embarrassing news — witness 365 Main’s disappearing of the press release that day touting a customer’s two years of continuous uptime at the same facility — but doing so is counter-productive. Customers will post on their ‘blogs and forums, and if you’re especially unfortunate, Valleywag will publish something with their usual level of fact-checking. The only way to maintain your credibility is to be the authoritative source by openly sharing and publicising the details on the home page, corporate ‘blogs, etc.
And to veer off in a different direction, I’m shocked by the number of “Web 2.0″ firms knocked completely offline by 365 Main’s power problem. SixApart is well-funded, from outward appearances seems to be doing quite well, and knows first-hand that data center power outages will happen! I said last month that power loss is the greatest threat to your servers and 365 Main’s incident is yet-another example. If you aren’t distributed it will eventually happen to you.
* I pay for this server at the old Interland employee discount rate ($50 off) with an additional discount for annual pre-payment. Peer 1 now offers a similar server to employees for free but I’ve declined to have mine converted.
