Moving to F5 LTM virtual appliance from physical appliances

The same test plan you used when you implemented the physical F5!

HA! I will accept that as comical bias.

Ok . . . access is different, now I have direct access into the box. Diagnostics are different; resources are not fixed. Bandwidth is different: now I have 8 physical interfaces on some boxes, but on the virtual box I will have a 20GB interface that I will share with 15 other blades. I proxy SSL now on a dedicated chip on dedicated hardware, but the virtual appliance will be using the VMware CPUs for SSL offload.

That's just the stuff I can think of . . . what else am I missing?

By the way, you said exactly the same thing our virtualization team said. "No difference."

Thanks,
Steve

Your test plan service should be the same irrelevant of the platform

If it fails test plan it FAILS!

The hardware may be different that's a given.

And that may be a concern but if your test plan is written correctly the test plan could fail

While I respect your observation that "a test plan is a test plan" I have to respectfully (as much respect as you can demonstrate on a text based forum) disagree. The vendor supplies performance and throughput expectations for the hardware appliance, but does not supply the same information for the virtual appliance because it is dependent on the underlying hardware. So I have to request resources from my enterprise Virtualization Consultant who will tell me emphatically that the vendor overstates resource requirements and will then give me what he thinks I need, and I will have to run performance tests with what I am given. I also have to support a wide range of applications, some of them in protected, isolated tiers that run weaker ciphers that are handled ok by a hardware appliance, but will suck the life out of a server CPU.

Thanks,
Steve

What does the specs of hardware versus virtual compare?

Thanks for the response.

I think I could have been a little more explicit with my question, but I feared that would put off respondents. Here's why: I am not really looking for a text book comparison of virtual vs physical because I don't live in a text book world. Fact is, during some testing, one of the VMware techs bounced my virtual instance without trying to take a snapshot when the "tool" he was using said my instance was in trouble. You would be right to observe that was improper behavior, and I have said as much to his manager. And you are also on the right track, though it smacked sadly of a little too much bias, when you say that my nose is put out of joint because, etc. I would phrase it this way: I have tools and experience with the physical appliance along with direct control over the box. Now some of my tools have been obviated and I am getting push back from the virtual team because I want access to some of the tools they use to monitor the health of my virtual appliance. And, on investigation, by my way of thinking, they are sadly under-tooled for the environment they are managing. And last, any comment I make about concern over performance seems to send them into a defensive tizzy as if I were accusing them of technical incompetence.

So I was looking for information on the operational realities surrounding migrating from phys to virt and I seem to be hitting the same wall. "There's no difference" and that's flatly and demonstrably inaccurate. Just pump 10Gb of UDP at wire speed toward a virtual box and see how well it behaves as compared to a physical box if you are still of the opinion that there's no difference.

Honestly hope I did not offend.

Thanks for your words.

Steve

Thanks!

Not offended and you are Correct but what you need to do is prove virtual is WRONG for your environment!

Also too add we see the deskilling in IT every week because management has given the VSphere Boys Management on new enterprise services!

And they vSphere Boys cannot manage or administrate!

We would not recommend snapshots anyway

Good Luck let me know how it proceeds