Will Windows 2012 kill the BDR? In the last few years Managed Service Providers (MSPs) have embraced Backup and Disaster Recovery (BDR) appliances as the solution of choice for managed backup. A BDR is usually defined as an appliance that includes backup software, offsite replication, and the ability to spin up virtual machines in a disaster. An MSP recently emailed me saying he felt BDRs were too expensive and didn’t make sense for a small business. His reasoning: Many SMBs have limited budgets and can tolerate a day or two of downtime. Spinning up virtual machines on a BDR can present challenges, requiring hours of technician time that might better be spent on getting the primary server back online. In addition, the BDR is less powerful than the primary server so it can be painfully slow for end users.
There are some new features in Windows Server 2012 that make me wonder whether purpose- built BDRs and even third-party backup software will play smaller roles in the future. For backup and disaster recovery, consider using two physical 2012 servers. They don’t need to be identical hardware. Each would run Hyper-V and have inexpensive local storage (no SAN required).
Using the new Microsoft Hyper-V replica software, virtual machines can be replicated between two (or more) physical machines. Hyper-V uses an asynchronous replication mechanism that has been designed for high latency and low bandwidth internet connections. It does this without taking all of the available bandwidth, which would impact normal email or browsing activity. One of replica machines could be offsite. I envision customers that will have a lower cost secondary server onsite taking the place of the BDR and a third server or cloud storage taking the place of today’s expensive BDR data centers. It won’t be long before established virtual hosting providers start advertising Windows 2012 VMs running replica for a monthly fee.
Microsoft’s replica is pretty clever software. Once enabled, the source host starts to maintain a HRL (Hyper-V Replica Log file) for the virtual machine. Every write by the primary Virtual Machine (VM) will also write to the HRL. Depending on bandwidth this log file is replayed to the replica VM every five minutes. The HRL mechanism replays the log file in reverse order allowing it to store the latest writes first. In other words, it is asynchronous (able to deal with long distances and high latency) and it replicates just the changes. Assuming they have the bandwidth to maintain changed files, running replica means the customer would lose up to 10 minutes of data worst case. It also will retain a configurable number of historical copies (versioning) on the destination side.
As a vendor of a roll-your-own BDR with removable drives (The BNAS), I’m sure we’ll want to incorporate Windows 2012 as soon as possible. Having built-in removable drives will provide a secondary backup mechanism as well as providing for “seed” backups needed for large data sets that take too long to copy over the Internet. Once the virtual machine is seeded, the replica engine will be able to keep it up to date in most installations. How well replica handles open files, AD, Exchange, SQL, and other tricky programs will determine whether MSPs use it as a replacement for backup software or just for replication and disaster recovery. If you try it, please drop me a line, and let me know how it works for you.
This article originally posted with SMBNation