Building an elastic environment requires elastic infrastructure

Ah, no, we won't likely be on the same page.

 

 

We haven't agreed on performance, reliability, or security at this point - only the suitability of traffic managers as virtual appliances for testing, development, and training.

 

 

You're not comparing MTBF of the solutions, nor are you including MTBF of each component required in a virtual appliance - OS, VM, and solution. You're not taking into consideration that the security of a virtual appliance relies upon not only the solution's security, but the security of the VM, the hypervisor, the OS, and all related software executing on that server. And you're not taking into consideration performance as it relates to throughput - hardware and software have vastly different throughput capabilities, and software is further limited by its reliance on the multiple layers of software required.

 

 

In my long years evaluating application delivery solutions for Network Computing I have never run across a software solution capable of matching the throughput and performance of a hardware solution. If you distribute the software across multiple instances, you might be able to match the performance, but then your CapEx story goes out the window as each instance requires more hardware and more maintenance and increases the chances of a failure.