For those of you who don't know, WireSpring is located in Ft. Lauderdale, FL, roughly half-way between Miami (hurricane Andrew, '92), and West Palm Beach (hurricane Frances, '04), and across the state from what's left of Punta Gorda (hurricane Charley, '04). Needless to say, disaster recovery and preparedness is something that we take very seriously.
A
few days before Frances began her slow pummeling of the Florida
coastline, we had a chance to test out our plans: phone lines were
re-routed, the computer services that could be were offloaded onto
other, off-site servers, our power conditioning equipment was tested,
and anything fragile or electronic was unplugged, wrapped in plastic,
and put in a closed, windowless room. While all of this was
irritating and time-consuming, never once did I think about our
critical services, like our CRM system, our live chat help system, or,
most importantly, our ClientCenter kiosk and digital sign remote management platform.
That's because the computers that run these services aren't located in
Ft. Lauderdale. In fact, they're not even located in
Florida. We host our services at a number of different US
locations near Tier-1 network access points (NAPs). Though South
Florida does have a Tier-1 NAP and a number of very impressive
datacenters built to withstand Category-5 hurricanes and severe
flooding, we felt that it would be best to keep our data in multiple
disparate locations in less meteorologically active areas. Thus,
our live chat comes from Washington, our CRM from California, and
ClientCenter from Virginia and Texas.
Of course, we also use high quality parts (I'd hate to say one thing
but do another :). Thus, our servers use actual server-grade
parts, filtered power, redundant power supplies and fans, and hard
drive RAID arrays. Finally, our IT folks are available 24 hours a
day, both on-site and remotely, to fix any other problems that might
crop up. For you Windows sysadmins that shudder at the thought of
remote management, remember, we're a Linux shop. Linux, like most
other UNIXy systems, was basically built to be remotely managed.
A good sysadmin can easily remotely manage 50-75 properly configured
servers, and actually we gain certain economies of scale as we add more
servers to our clusters.
I bring up this last point to
illustrate what it takes to run a highly reliable kiosk or digital
signage network. Not only do you need to make a capital
investment in server hardware and software, you also need to provide
connectivity, service and support to keep it all running. Factor
in things like operating system and other software upgrades, and you're
looking at a pretty complex undertaking.
WireSpring's critics
like to point out that we won't sell our ClientCenter software as a
boxed product -- we only offer it as a monthly service. We feel
that given the amount of time and money invested in our infrastructure,
and the amazing amount of knowledge amassed by our technical team,
nobody else could maintain the same security and reliability of our
systems as we can. To those people who still think they should be
able to buy ClientCenter, I say this: when you order cable or
satellite TV, do you demand to own the transponders, cable lines,
satellites and other equipment? When you pay your electric or
water bills every month, do you grumble because you don't own the
infrastructure? Do you have the skills, know-how or desire to
operate your own telecom system so you can run phone and Internet
services directly?
Probably not.
Even if you do, factor
in the cost and complexity of running the systems 24 hours a day,
monitored for any problems or interruptions, with tech support at the
ready, and a full disaster preparedness plan in place, and you'll see
why we can make this a full time job :)