|
June 4 2009 — Servers crash. After all, they are just a lot of circuits that can fry, a power supply that can burn out, a fan that can stop working, and a disk drive that can fail. They crash a lot less frequently than they used to. And, while twenty years ago crashing hard drives were not-uncommon events, they are rare today. Still, it does happen. One solution is spare equipment. I worked on a project for Citicorp where every server had two external disk arrays that mirrored each other and there was a standby server that could replace the one that crashed. They could have a failed server back in production in 15 minutes. But a small business or non-profit doesn't have the bucks to spend on standby servers and redundant disk arrays. So what can you do?
Well, fortunately, technology has come up with some good alternatives. And while they were first developed for large organizations with big bucks (and big losses from downtime), there are some reasonably priced solutions that fit smaller organizations. These solutions allow you to get back in business quickly. How does this work?
Let's start with an example: Imagine you run a commercial bakery that takes orders all day. Normally, production will start early in the evening after all the orders are in and your production requirements have been printed out. Now, suppose your server has a hardware crash at 3:00 PM. In the old days, you would have to get the server repaired -- next day -- reinstall all your server software and then restore the backup sometime early next afternoon, if you were lucky. In the meantime, what about the day's orders and production? You can imagine the potential financial losses, staff working late into the evening, production chaos, late deliveries, and the customer service nightmare.
So, is there something you can do, short of having spares of everything, including a standby server? Yes, and there are a couple of parts to this answer: First, there are backup tools -- such as Symantec's Backup Exec System Recovery -- that backs  up what you are doing to an external hard disk on an on-going basis. It can also restore the backup to "bare metal" -- that is, can recreate your server from scratch on a "new" machine. The second piece is setting up a "virtual machine" on another computer, using a product like VMWare or Microsoft's Virtual PC or Virtual Server (now Hyper V). As far as the world is concerned, the "virtual server" is exactly like a hardware server, except it lives in another computer. This allows us to take the backup we have been making on the external hard drive and restore it to the virtual server hosted on the other computer. After restoration, you can start up the virtual computer and you are back in business.
Ideally the machine that hosts the virtual server is another server with lots of memory (minimum 4 GB), but for really small shops this can work with a good workstation, like a Dell Optiplex or similar computer (also with lots of memory). The virtual server won't be as fast as the original, but you will be able to work until the real server is repaired.
As a third component, I recommend either a tape or removable disk cartridge device. Make a copy of the backup weekly and take it off-site for disaster recovery purposes. I have heard from several sources that 75-80 percent of companies that have fires and no off-site backups don't reopen. I suggest a three tape or cartridge rotation.
Trackback(0)
 |
Keep up. Follow us on Twitter 
Gary's Blog 
TrackBack URI for this entry
Subscribe to this comment's feed