I've spent a lot of time developing disaster recovery plans for mid-size companies with 200-2000 users and 10-100 servers. These companies don't have the multiple data centers and dark fibre that the big boys use to enable array-to-array replication. They make do with servers at a co-lo site and use host or application replication. Double-Take Cloud has extended host-based replication and application recovery to Amazon's public cloud.
I've considered using EC2 as a disaster recovery solution in the past, but the math didn't add up. First, you'd have to buy replication and application failover software like Double-Take, SteelEye or CA XOsoft, for both the primary server on your premises and the EC2 server. Then you would have to pay Amazon to keep a server big enough to run your application 24-7 when the disaster strikes, so it can run the replication data catcher. My math says that adds up to $5000 plus $350/mo for an EC2 "Large" Windows image, not including storage and data transfer costs.
Double-Take Cloud simplifies the process. You install Double-Take Cloud's back-up agent, which uses a Windows file system filter to trap data changes, and then you set up an account. Double -Take Cloud replicates to a Double-Take Cloud repository EC2 instance, which can accept replication from multiple source machines. When a source server fails, or a flying saucer comes down and squashes your office, you can quickly spin up an EC2 machine that mounts the Windows image of your source server. Double-Take's recovery process swaps the IP addresses and gets the server running.
The system uses OpenVPN to encrypt the data in flight and when the application is running on theEC2 server can replicate data back to the original source, or a new server, so you can return to normal operations as easily as possible.
Double-Take is charging just $150/mo plus Amazon storage charges which makes Double-Take Cloud look like a bargain to me.
