Very interesting, and thanks for coming through with more detail so quickly. I look forward to hearing about the fundamental reasons you think you are seeing this performance multiplier on these systems.

I'd also be interested in hearing about how this setup compares when the Amazon EC2 filesystem is stripped over multiple EBS virtual disks (exactly how much work and expense would it take to get EC2 to a similar level of performance, if it's even possible?), but given that you're providing a turn-key accelerator solution, I think this is a very fair and very impressive comparison.

Rod, thanks for the response. I haven't been fair, as this is a really good looking offering regardless of how much better the performance is than an EC2 setup with similar software.

I know that Joyent is creating more of these types of bundled offerings, which I think is pretty great. I look forward to seeing what is coming up in the pipeline (and those benchmark details ;-).

I agree, this is not a helpful metric. It should either not be offered or should be backed up with data. If the explanation of the benchmarking process is not available, then why not wait to make the claim?

Based on the marketing page it looks like you may have been using EC2 local storage instead of EBS attached drives. It is my understanding that the EC2 local storage is significantly slower than EBS and that EBS is more similar to Joyent's storage services. Hopefully the benchmark is an apples-to-apples comparison, but without data I don't see why I should believe it is.