The Apollo Property Services Group, which builds and maintains homes and schools, has 100 offices across the UK. It depends on its wide area network (WAN) for access to SharePoint and collaborative project applications. A WAN disaster recovery plan to ensure the availability of these applications and its key data is critical.
Mark Bowell, infrastructure and network manager for The Apollo Group, runs the company’s business continuity and disaster recovery plan to keep services up and running for users.
“Our concerns were ease of implementation and ease of use. In case of disaster, the last thing you want to be doing is ‘now hit this, now do that.’ You just want a nice, easy solution,” said Bowell.
Disaster recovery for additional applications
The enterprise had a data duplication solution, Double-Take Availability from Vision Solutions, which was implemented initially to protect Apollo Group’s file servers, followed by Microsoft Exchange and Microsoft Dynamics NAV enterprise resource planning software. Data is replicated from the production servers across the network to secondary servers.
“We introduced Microsoft Exchange 2010, and we needed to make a decision about Vision Solutions or rely on something else. We tested a couple of other products, but found them overly complicated and reliant on specialist configuration,” said Bowell.
It also became critical to have a disaster recovery plan for the enterprise’s SharePoint application. This is hosted within VMware with images taken, which provides it with some high-availability capability. However, the database that is used to support the service is too large and too critical to be hosted on a virtual machine, and has its own dedicated server.
Testing WAN disaster recovery
The solution has performed well in the face of two real-life challenges. Bowell does not rest on his laurels, however; the system undergoes continual testing to ensure it will work as planned and the network infrastructure will support it.
“We have implemented and tested it two ways for failover and failback. If the primary has a problem, people don’t even know about it. It takes seconds. You can then fail it over, seamless to the users, which is great. The pressure is off of IT,” he said.
“We then fix the problem on the original, which means we had to fail it back so you don’t have two machines in the same location," Bowell explained. "You potentially stress the WAN links, but in this case there wasn’t any of that.”
“The other way we implement it is to force a failover,” he said. His team runs a test disaster at least once every two months.
Bowell recently tested out a disaster scenario: “If we had to fail over completely and we were stressing the WAN, would all our clients be able to connect seamlessly? Over Christmas we tested the WAN to make sure it was resilient enough to provide our clients with ease of access. Protection of the WAN comes down to network. We ensure our network is monitored, bandwidth control is implemented, [and we have] segregation of traffic to make sure people are not getting unnecessary broadcast storms that would certainly not help.”
Bowell’s latest innovation is to switch disaster recovery capabilities to the virtual space. “We are thinking about going from physical to virtual. We have tested out Vision Solutions going from physical to virtual because we have a virtual infrastructure,” said Bowell.
“That is on the horizon," Bowell promised. "We have tested it, and we will revisit it at some point in the next six months.”
--Tracey Caldwell is a professional freelance business technology writer.