Hello there. I’m back to continue my blog series on considerations for DR. Here are a two more questions I am frequently asked by customers and partners, as well as my answers:
Q: Alternate locations, do I need a live (hot) DR site or copies of data in a static (cold) location (e.g. Iron Mountain, Sungard)?
A: The business criticality of the systems you are using will determine what provisions you make to supply the business with the relevant services and data during a disaster. This, in turn, will allow the scoping, sizing and type of infrastructure you may need to use to deliver the service level agreement back to the business. Core business critical services can often be delivered back to the business in times of disaster by having copies of the services and data at alternate locations outside of the main business location.
An interesting idea on this that’s gaining popularity is to actually have your normal production systems offsite in a secure managed data centre, while the DR location is actually back at the company office.
This type of setup can offer some benefits, your core business critical systems are now running in a controlled environment, with good resilient services like power and air conditioning. If for whatever reason the data centre becomes unavailable, your manpower needed to implement your DR strategy is now actually available in your office—not in a remote located data centre with restricted access.
If alternate locations are a must for DR then you need to consider how you keep up-to-date datasets available and protected at both locations. For the data that’s not so mission critical but is still vital to the business, we can look at technologies that allow us to store more data for longer periods of time, allowing for faster recovery.
When working with data protection in environments like data centres that charge for power, cooling and rack space, technologies for data protection like deduplication start to pay benefits. Not only does the technology allow you to store more data for longer on smaller disk based platforms, it can reduce the amount of visits required to change traditional based tape backup media. But that’s not to say that tape doesn’t need to play a part in your strategy; it still has the innate capability to offer easy offsite copies of data that can be kept in secure controlled environment.
Q: I have multiple locations. How much data can I send offsite? And will my data be any good when it gets there?
A: If an alternate location for DR is a required option for DR, we can use our data classification to determine what we really need to get to the DR site. There are many technologies available that help achieve the movement of data to alternate locations. If we know what data we need to move and have settled on the RTO and RPO requirements, you should end up with a quantity and time requirement to move that data.
This, in turn, will show the inevitable need for bandwidth. Bandwidth is an interesting thing in IT; it seems to be the only item where its cost is not inversely proportional to its speed! Everything else over the years has got a lot faster and a lot, lot cheaper!
This sometimes becomes a stalling point for DR and prevents businesses from delivering what they need and drives them in to the “It'll never happen to us anyway” mindset.
While looking at DR strategies and data movement one of the more popular methods—especially when alternate data centres are going to be used—is virtualisation. The use of virtualisation technologies allows the possibility of deploying many standby servers in a smaller hardware footprint than a traditional physical solution.
So now we begin to see cost reductions taking place around the hardware. But the real benefits of virtualisation start when it’s deployed across multiple locations. With technologies such as VMware’s Site Recovery Manager making your core systems available becomes easier to manage in terms of disaster recovery. These solutions usually take care of managing the virtual machines that provide the service, but we also need to consider the ever-changing datasets that sit behind the service application that are not covered by this kind of technology. Protection of the virtual machines themselves is also another aspect. Just because you’re replicating them doesn’t mean you should forgo any kind of protection solution. There is always a need for data protection in virtual environments. Part of any deliverable DR solution that crosses multiple sites will inevitably have to deal with the replication of the changing datasets.
And since we’ve already discussed the needs surrounding bandwidth, RTO and RPO, we should also consider the data itself. Content is King and key to recovering business activities. Servers, VM's, applications can be recreated; content cannot unless you have a good backup with consistent datasets.
There are many products that will enable data replication, some at a block level on disk-based SANs and some at the host and application levels. These replication solutions are all well and good, but as with any replication technology, the old adage applies, “garbage in, garbage out”. Replication technologies don’t always give the ability to catch issues around data corruption or logical data deletion. Considering an ‘out of band’ solution to protect, replicate and recover the changing datasets allows for greater flexibility options when the time comes to recover the data when you really need to.
In my final blog on this topic, I’ll reveal the one thing you can do to ensure your DR strategy will work.
Go back to review Part 1
Continue on to Part 3