2017 Prediction: The Death of DR
Enterprises have spent billions of dollars and decades of time honing their Disaster Recovery (DR) plans and processes – 2017 will be the year that DR will die. How could such a central tenet of IT operations disappear? Simply put – it just doesn’t cut it anymore.
DR depends on having a second set of resources – equipment, data, people – ready to take over operations when your first location experiences a disaster. You maintain those systems miles away, to avoid a natural disaster from taking out both locations. You practice the transition of operations at least once a year, and you document every step needed to succeed in the transition. This mode is called active/passive or active/idle.
The problem with DR is that it has “disaster” as its foundation, from which you “recover” to a second alternative site. The idea of your business withstanding downtime – even short-lived downtime – doesn’t fly in today’s digital world.
So what will we do instead of DR? Continuous availability.
Designing for continuous availability means you’re running operations in multiple locations all in active mode. This mode is called active/active.
Continuous availability offers a broad set of advantages. The first is obvious – you’re continually operating rather than failing and then recovering. So your customers never see a loss of service.
Continuous availability also offers other advantages, too.
First, it’s more economical. Let’s take the simple case of two data centers. With DR, you build 100% of your capacity in your second data center. You need to, in case it ever has to take over – it has to offer the same horsepower as your primary site. But all that equipment, WAN connectivity, personnel at the ready – and you never use it! It sits idle, depreciating, and not servicing any of your customer needs.
With continuous availability, you design for a lower capacity – perhaps 70% in both location – but you’re running live operations in both locations simultaneously. You can do that because you never run your primary site at full load – if you hit 50% utilization, you increase capacity.
So at 70% in both, you’ve got 140% of capacity, and even if just one site has to handle all operations, it’s got the capacity to do so. So now you’ve spent 140% instead of 200%, so you’re ahead on equipment costs. You’re getting full utilization out of the equipment instead of having it sit dark.
Second, active/active operations increase performance. With services running in multiple locations, your customers will see a faster response time to their requests. Many organizations choose to run active/active operations quite far apart, so users on each side of the country can be serviced by a data center much closer to them. Avoiding the delays of traversing a long WAN link will boost throughput.
But the active/active operations needed to deliver continuous availability also brings its challenges. Building another set of servers, networks, and WAN links is fairly simple – having data straddling two locations is where things get sticky. You’ve got to invest in modern databases that support scale out and can have servers in multiple locations with replicated data. Then you’ve got to ensure your applications know how to talk to those databases. One trick is to leverage software like ScaleArc’s database load balancing software to enable your apps to read from multiple database servers as if they were one large single server.
Continuous availability, leveraging active/active operations, is essential to meeting today’s digital business demands. Your team will have new tricks to learn, but you’ll deliver the service level, strong economics, and high performance that define success in 2017.comments powered by Disqus