When Microsoft first showed off its disaster recovery service on Azure in 2014, it was called Hyper-V Recovery Manager and was an extension of a System Center tool for failing over Hyper-V virtual machines (VM) to another location, using the public cloud to coordinate testing and managing recovery between your data centers.
Today, Azure Site Recovery (ASR) lets you fail over into the cloud as well to another location, from physical and virtual servers, and it handles far more workloads than just Hyper-V and System Center Virtual Machine Manager (or even VMware and vSphere, which it also supports).
“We’re an application-centric disaster recovery solution,” Srinivasan Chandrasekar, director of program management for ASR and Azure Backup, tells CIO. “Getting your application to work is more complex than protecting one virtual machine. How do you protect your Active Directory infrastructure? How do you protect your VM infrastructure? If you’re running a three-tier web application with Oracle or SQL infrastructure in the back end, how do you protect that? How do you protect management infrastructure like System Center Operations Manager? How do you protect a SharePoint application?”
ASR promises to do all that and more. “If you replicate a virtual machine, in theory any workload should be supported. There’s no reason a line of business app built internally shouldn’t work, because we are app agnostic in the core technology,” says Chandrasekar. That includes apps that need Active Directory replication, specific IP addresses, load balancing and other infrastructure to be in place, not just simple VMs.
But because ASR focuses on protecting applications that have complex infrastructure rather than just backing up virtual machines, specific workloads have been tested and certified. Those cover Active Directory and DNS, web applications built on SQL and IIS, System Center Operations Manager, SharePoint, SAP, Exchange, Remote Desktop and VDI, Oracle, Dynamics AX, Dynamics CRM, the Windows File Server role, Linux servers, and now Citrix XenDesktop and XenApp (all of them running on VMware as well as Hyper-V).
“We have solutions that we have tested for various applications and topologies, which makes it easier for you to deploy this — and to check that it’s working for your environment,” Chandrasekar explains. So when you protect XenDesktop, that covers the Citrix Delivery Controller, the StoreFront Server, the XenApp Master Virtual Delivery Agent, the XenApp License Server, your Active Directory DNS Server and your SQL Database Server, and you can add your own automation scripts to the recovery plan.
That’s a server workload with the infrastructure that supports it, Chandrasekar notes, because Azure doesn’t currently support running Windows clients in VMs, but when the XenDesktop Essentials service arrives in Q1 2017, letting you run Windows 10 Enterprise E3 desktops on Azure, you should be able to protect that as well. “Whatever workloads Azure supports, it is our intention to support using Azure Site Recovery.”
What’s next for Azure Site Recovery
Given how large a part of IT budgets storage has become, interest in cloud backup is natural; Gartner predicts that the number of businesses doing cloud backup will double between 2016 and 2018. “Everyone is asking, ‘How do I back up everything to the cloud?’,” says Elden Christensen, Lead Program Manager for Microsoft’s Failover Clustering team, “and DR is following.”
Disaster recovery as a service (DRaaS) started out as something that appealed to small and midsize organizations that didn’t have the budget or the skills to manage failover sites, but as interest in hybrid cloud has grown, enterprises have also started turning to DRaaS.
Cloud DR can deal with the issues of cost and complexity that have held back DR adoption, says Chandrasekar. “DR as it has existed is complex and unreliable. How do we democratize DR? Backup is democratized; everyone backs up data and you don’t think twice about it, but to make sure that your applications are up and running in the event of a disaster, small or large, is very expensive, very complex — and you don’t know when you deploy a DR solution whether or not it’s going to work. With cloud, we have reliable technology that can reduce the cost of doing DR and make it simple to use and to deploy. Because cloud delivers DR as a service, you don’t have to deploy a management server, you don’t have to run the brains of DR on premise.”
While ASR offers disaster recovery from one data center to another, “more and more customers are realizing that it’s much more cost effective to put DR in the cloud,” Chandrasekar says. “Over the last six to eight to 12 months, we’ve seen that the interest in protecting to the cloud is far outpacing interest in protecting to on premise.”
Most customers start with a single workload. “It could be a line of business application built internally, it could be SharePoint or Exchange; they start with one and get a feel for the service, and then we see it expand to a wide variety of workloads.”
You can also use ASR as a migration tool to move your applications from your on-premise systems (or from AWS) to Azure in an orchestrated, tested manner; either straight away, or the next time that you have a hardware failure.
“We have customers using Azure Site Recovery who say, ‘I’m going to migrate to the cloud, so let me use this technology to replicate my applications and do testing because of its failover capabilities — and then I’m just going to switch over’,” says Chandrasekar. “And I know of a couple of fairly large customers who are protecting hundreds of virtual machines using disaster recovery to the cloud, and their stated position is ‘if I ever switch over [to the cloud] because something goes wrong, I’m not coming back’.”
To simplify that, the team is working on extending to Azure Site Recovery the Azure Import/Export Service, which lets you ship hard drives to Microsoft to have the contents copied into Azure blob storage rather than waiting for them to upload.
Other customers are interested in using ASR to protect their workloads in Azure. You can already migrate, replicate and failover VMs to another Azure region, but they don’t yet fail back to the original region. That’s also in development.
Azure Stack, Microsoft’s on-premise hybrid cloud system, will also use ASR, giving you disaster recovery to the public cloud or to another Azure Stack region that you’re running.
Also in the works is a way to protect VMs running on VMware to a Hyper-V system, which would also let you migrate from one hypervisor to another.
Room for improvement
Some limitations are down to Azure itself, like the one terabyte disk limit; “in future, that will no longer be an issue,” Chandrasekar promises. Also in the works is a way of tracking which workloads you’ve created recovery plans for and whether you’ve tested them. “We’re adding a health model in the product to make sure customers are aware when they're protected. It’s one thing to have the service running; it’s another to click a button and know that you’re protected.”
Another area for improvement is security, especially as applications like databases as well as files are now being targeted by ransomware. “It’s a question we get more on the Azure Backup site, where we’ve added capabilities to make sure they can protect themselves more, but we’ve also had conversations with some customers around ransomware and how Azure Site Recovery fits in and we’re looking at ways we can add more defense in depth for customers.”
This story, "What's next for Azure Site Recovery?" was originally published by CIO.