One of the most persistent and baseless misconceptions about private cloud is that you don’t need a disaster recovery (DR) plan for your business applications.
Yes, private clouds are built with high availability to maximize uptime. But that doesn’t mean they’re immune to failure.
Whether the private cloud you’re using is hosted in your data centre or at a service provider offering you infrastructure as a service (IaaS), platform as a service (PaaS), or completely managed services, there’s a chance each can fail. The tipping point might come in the form of a natural disaster like a hurricane, flood, or snowstorm; a cybersecurity threat like ransomware or malware; or even an equipment failure.
No matter what your private cloud infrastructure looks like, you need a DR plan in place to guard against these risks. Here’s what that looks like.
Essential elements of disaster recovery for private cloud
Beyond the typical requirements of an effective DR plan, private cloud DR can present additional challenges. To address them, DR for the private cloud should have:
1. Seamless network transition
Regardless of whether you’re operating from your private cloud or your DR environment, the user experience should be completely seamless. No matter where an IP address travels, users shouldn’t be able to tell if you’re running at a different site.
What makes this so difficult to achieve is networking. Without the proper connectivity, your recovery environment might not be able to deliver the same application and workload performance your users expect.
With Sungard AS’ Private Cloud, that’s no longer an issue. By partnering with Megaport, Sungard AS ensures customers have access to a globally available Software Defined Network (SDN) that delivers on-demand, scalable, and secure connectivity to leading cloud providers and top-tier data centers around the world.
2. Easily replicable backups
If your backups are in your main production environment, but you start running out of your recovery site, how do you recover these backups at your new location?
Like with the network transition, the goal here is that users will never know the difference between production and recovery. The ideal DR plan for private cloud enables replication of your backup data in your recovery site so you don’t miss a beat.
3. Granularity of control
You should have the ability to declare a disaster on your own, rather than be at the mercy of your provider. But it goes further than that. You should also be looking for the granularity within this control.
You should be able to declare a disaster without your entire site going down. For example, if a single database is corrupted, it’s important that you can declare a disaster for just that database. Or, if you’re running 100 virtual machines (VMs) and declare a disaster but want 20 of those VMs to go to a particular DR site, you should have that capability as well.
Issues to watch out for
Some of the features and capabilities you have in private cloud don’t necessarily match those in your recovery environment. You also need to keep in mind exactly what services your private cloud provider offers and what you might be missing after you fail over.
Think about these three elements based on your dependency on the infrastructure you’ve selected:
1. Post-recovery management
Many private cloud providers are more than capable of taking care of your recovery. However, it’s the “Day 2” management – what happens after you recover – that you need to be concerned about.
Is your private cloud vendor capable of handling your post-recovery management? Are you contractually covered to run at a DR site and receive the same service from all your partners? What if you recover in a different platform? Be sure you know where your provider’s capabilities end and what else you’ll need in place to manage your recovery environment.
2. Available controls
The controls you have on your private cloud may be different than the controls you have on your recovery target.
For example, running your private cloud in a provider’s data center may give you complete control of the stack. However, if you recover to a public cloud, you only have control from the operating system and above, meaning you’ve lost two layers of control.
So, make sure you know what controls are available to you.
3. Data protection
Do you know how you’re going to continue protecting your data if you start running at a DR site? Do your backup needs change, or do you need the same backup service at your DR site? Can your backup software support a changing backup target to an alternate storage? How would you repatriate the backups post-failback?
A private cloud DR plan is essential
Private clouds are built with high availability in mind. But that doesn’t mean you can afford to overlook DR.
Oftentimes, incidents such as infrastructure failures, regional disasters, cyberattacks, or even fires – like the one that consumed OVHcloud's SBG2 data center in Strasbourg, France – occur outside your private cloud provider’s control. Each layer of your private cloud stack is subject to various elements of failure, making DR essential.
You’ll need a DR plan tailored to your specific private cloud infrastructure, one that factors in how you can minimize disruption for users and ensure you have every element covered – no matter where you recover.