Why have so many organisations historically failed to prepare properly for IT disasters? Well, the fact is that IT disaster recovery (DR) is complicated. Doing it well requires a lot of planning, and you’ll have to keep on updating the plan if it isn’t going to be totally useless by the time you want it. Testing the plan properly is almost never done, as it’s got a tendency to be ruinously time-consuming, says Plan B DR’s Tim Dunger.
Traditionally IT disaster recovery (DR) solutions have been complicated and expensive, or cheap and ineffective. This has left many with either inadequate provisions or worse, expensive provisions that fall short of protecting the business. Also, if the worst does happen, you need to think if it is best to be solely reliant on your own IT team. They will probably be up to their eyes trying to deal with the cause of the disaster, or worst of all, may actually be caught up in the disaster itself. So what are the options for protecting critical IT systems for your organisation?
Online backup is basically no different to a local backup but has some definite advantages. Not having to keep moving tapes around is a step forward and you can make an argument that your data is more secure in encrypted form than in plaintext on tape in somebody else’s offsite storage. But, depending on your online backup solution, if you start thinking about recovery, it’s not all roses.
If your online backup product doesn’t keep a local copy of all of your data, how long would it take you to download a complete set of all the data on your biggest server, using a sensible proportion of your Internet connectivity? And if you’ve got a better online backup product that does keep a local copy, what happens if it’s your entire building and IT systems that are out of action?
Have a plan to replace and rebuild
This is probably the most popular form of disaster recovery provision in the mid-market. I’d like to suggest that unless you’ve got a detailed, documented plan of how you’d replace and rebuild your critical systems, then your backups, even if they’re great, have rather missed the point. After all, what really matters to you is not that your data is ‘safe’ but that it’s useful. If you can’t interact with it, then it’s not doing any useful work for you. If we just take a simple scenario for bringing a few systems back, including their backup server, you might have a process that looks like:
- Make sure you’ve got all the kit you need. Remember that not all kit is immediately available (Dell build machines to order, Cisco devices have a habit of not being in the channel when you want them, and tape drives and libraries aren’t always easy to acquire at the drop of a hat)
2. Retrieve operating systems media (was it destroyed?)
3. Locate activation key for operating systems (where is this?)
4. Re-install the operating systems
5. Patch the operating systems (assuming Internet connectivity is good)
6. Locate backup software media
7. Locate activation key for backup software
8. Load backup software
9. Insert first tape and integrate tape database back into the backup software database
10. Identify and retrieve all tapes in cycle
11. Start full restoration
So… we’re eleven steps in, five of those steps require you to either have a copy of something, or to get bits from somebody else and we’ve not actually got a single machine back up and running properly yet. The fact remains that rebuilding your systems is a highly complex and time consuming process—it can easily take up to 7 days! Unless you’ve done it from a realistic starting point a couple of times, it is unrealistic to expect it to work when you need it.
Warm and hot standby
Finally, we start to get to solutions that actually stand a reasonable chance of delivering quickly when you need them. ‘Warm’ and ‘Hot’ standby solutions take a starting point that you need to have properly provisioned, permanently available, configured, and working systems in another location, ready to take on your workloads?as and when you need them.
Where they differ is in terms of the currency of the data on them. Typically, a ‘warm’ system will be fed data periodically, so should you need to press them into service, you’ll see your data go back in time somewhat. A ‘hot’ standby system is continually replicating data, so you should see minimal data loss when you need it.
The downside of both of these approaches is the same?cost. Depending on your solution, the level of importance of your data and the importance of keeping up to date with every last transaction, prices can range from the inconveniently pricey to the truly legendary. Also, bear in mind that these solutions aren’t a replacement for a good backup regime?they’re an addition to them. So you need to pay attention to your backups, as before.
The promise of virtualisation
Virtualisation as the buzzword of the moment has a lot to live up to, but when it comes to disaster recovery, it really does show some promise. In fact it should be able to solve both the drawbacks of traditional disaster recovery approaches by dramatically reducing costs for vastly more effective, fast recovery solutions.
Virtualisation’s promise comes from its ability to provide a consistent computing environment regardless of the underlying hardware. This means that a system image can be made to work on any hardware, and because of the consolidation advantages of virtualisation, multiple workloads can be run on single physical machines.
Virtualisation can deliver significant advantages for disaster recovery because it can dramatically reduce the hardware costs for the recovery platform. As a result you can now purchase products that would allow you to make a virtual warm standby solution from a physical live server.