Last week I got to go on a tour of a data center—a place designed to house huge arrays of servers securely.
I went on this tour because I wanted to know more about how ASPextra.net, the data center’s largest client, works. The concept of an ASP (Application Service Provider) interested me when I first heard of it. When you use an ASP, instead of buying all the extra hardware and software and installing and maintaining it yourself—and then having to worry about things like synchronizing your home, laptop, and office e-mail and other data—you ‘rent’ all those things from an Application Service Provider. This is not necessarily inexpensive, but can potentially pay off over time for small businesses and particularly for people who travel a lot. (To find out more about ASPs in general, see the excellent article by How Stuff Works.)
But before I started recommending ASPextra.net (that’s their web address as well as their name) to clients, I wanted to find out whether they were really providing the kind of security and ease of use that they claimed. And, naturally, I was hunting for information on the subject of backups.
So there we were down at ColoServe in San Francisco. In addition to needing to hand over my driver’s license in exhange for a pass to get in, I certainly saw plenty of evidence of physical security: the computers are in locked cabinets in locked rooms, the building has its own generator to provide power in case of rolling blackouts and an advanced fire control system, not to mention the air conditioning that comes up through grids in the floor to keep disks from overheating.
Why is air conditioning part of server safety? An overheated disk or processor is more likely to fail—if your computer gets too hot, it will crash. Don’t ignore cooling fan error signals, and make sure you dust the vent at the back (or bottom, with most laptops) of your machine frequently.
Then there’s security in the sense of having a network connection that’s always on: the “pipes” bringing data to the entire West Coast runs directly underneath the data center, meaning that for anything to interrupt internet service to ASPextra’s servers, it would have to be something that would take down the net for the whole state. There were redunant network cables on all the machines, so in case one failed, information would keep flowing.
We also got a demonstration of how the ASPextra spam and virus filtering system works to keep most viruses and spam from ever reaching their clients’ computers. It all looked pretty solid.
But backups, I said. How did they handle backing up their clients’ data (and software)?
The answer was “Actually, we typically use a RAID system on the client servers with dual or sometimes quad internal HDs. If the client wants a dedicated backup server, that is no problem, but that does cost extra.”
I’ve heard the term RAID a million times, but I had to go look it up in the Webopedia to be sure what that meant. RAID stands for Redundant Array of Independent Disks. There’s that word “redundancy” again. While it’s something you want to avoid in your writing, redundancy is something you want to achieve when it comes to your data. It’s like carrying a spare tire.
In a RAID, if one disk fails, another can replace it, thanks to “disk mirroring”—in this case not quite the same thing as the disk imaging done by programs like Norton Ghost, because the duplicate information is written simultaneously, all the time. (Incidentally, buying your own RAID would run you anywhere from about $2500 to $20,000, according to a C|NET search.)
Asked what the failure rate of the disks was, our guide said that only one disk had failed in 15 months. For drives in constant use, that’s an impressively low failure rate. The Ur-Guru (who was on the tour with me) has had many more disks than that fail in the past two years—probably due to inadequate cooling.
If you consider an ASP, make sure they can provide all of that. In the meantime, keep your computers cool and keep backing up your data.
Until next week,