It’s a really sad thing when a customer gets far down the path of a project and finds out it is going to fail. I can’t tell you the number of times that I’ve seen this happen. A customer has a problem to solve, they ask some questions of vendors, and maybe do a pilot program with one of the vendors. But it’s not until they go full production do they really find out what the product is like. Then sometimes it’s really hard to go back. New technology or products are where this happens most often, and it is because the right questions have not even been thought of yet.
One of the best examples I’ve seen of this was a customer that implemented an early model of a deduplication appliance. They tested that the device could ingest backups at a rate of 400 MB per second, and they tested that the backups written to the device were able to successfully restore their source data. They even tested an export of the data to tape. What they didn’t test was a significant number of simultaneous restores or simultaneous exports to tape. That’s when they found out that the device that could ingest data at 400 MB per second capped out at 40 MB per second during exports or larger stores. The device was already in production and paid for. It would’ve been a resume producing event (RPE) for the admin in question to reveal to his boss that the Emperor had no clothes.
The Right Question for DRaaS Vendors
So make sure when you’re examining DRaaS vendors, you at least ask all of the appropriate questions. The questions that you already know are about recovery time objective (RTO) and recovery point objective (RPO). How fast will the system allow me to recover, and how much data will I lose when I recover? But you need to look at these questions in a slightly different manner than you might be used to.
The ability of a DRaaS vendor to successfully satisfy your RTO and RPO requirements depend as much on environmental factors as they do the capabilities of the product. For example, you could install the best replication product in the world on a T-1 line, and it’s going to get nowhere fast. Or perhaps you have the fastest network bandwidth possible, but your data center generates an exorbitant amount of data. I know of one client that is actually a bandwidth provider that creates more data every day than they could successfully replicate – and they own the bandwidth.
You also need to understand the capabilities of the product and the infrastructure upon which the product will run. Here’s the question: What will the performance of my recovered data center be like? Will it be able to take the place of my entire data center in case of disaster? If you’re going to recover into the cloud, you might be there a while. Make sure the cloud provider and application that is providing the recovery is capable of supporting your entire data center. We dive deep with Infrascale during our on demand webinar.