1) You need client virtualization in every device. The Trango acquisition gets all of the embedded stuff. VMware already has good partnerships with the thin client guys. The desktops, laptops, and tablets are all taken care of with VMware Workstation/ACE/Player/Fusion.

2) You need to get rid of the OS in the VM. There’s just too much overhead there and it’s really not necessary. The main purpose of the OS is to abstract the hardware from the application and then to schedule out resources. Guess what? That’s what the “hypervisor” or virtualization layer does. All you need now is some JeOS like you said that is highly tuned for that specific app and you’re in business. Look at the BEA appliance as an example. There’s no good modern OS that’s specifically tuned for Java. Java always does funky stuff with memory. However, get a VM, write your own microkernel that’s highly tuned for Java, and presto – you now have a virtual appliance that runs with 50% less memory and 160% the performance of the physical environment. Pretty nice. That same example can go to other apps as well. There are already vendors such as rPath making JeOS for people to use. Now with OVF (open virtualization format) you can truly build the virtual appliance once and have it run on any virtualization layer.

3) The last trick is out of the hands of the virtualization vendors. You need to slim down the apps. Sorry to say, but I can’t run Exchange on my iPhone. I don’t think I’d really want to so that’s a bad example. However, I wouldn’t want to run MS Office either. I guess I stretched it too far saying that you could write the app once and move it from datacenter to iPhone. However, I do see a day when you could get pretty close. Client side apps are getting thinner and thinner. The fat ones are moving more into VDI environments or being published with Terminal Services or XenApp. The thin ones really could run on your desktop or your iPhone. The fat ones you’ll get to over the Internet just like you do from your laptop. Of course with all of this there’s the whole usability thing. Even if I did get Excel to run on my iPhone I’m pretty sure I’m not going to use it for anything but viewing.

There’s still a long way to go with all of this. Cloud, ubiquitous client virtualization, etc all will mature over the next 3 years in my opinion. They’ll get shaped by people like yourself coming back to us and saying, “hey, idiots…did you think about XYZ”. That feedback is critical to VMware’s success so keep it coming. Together we can define this and get it to work.

There are two aspects to this strategy that I struggle with that I would be interested in your commentary on.

1) The form factor and resource requirements of the two devices are totally different. This I think would make a mobile operating system totally unsuitable for use on a desktop and laptop and likewise desktop OS’s would be totally unsuitable for mobile devices. There are certainly advantages in a standardized mobile hypervisor. However, rather than one truly universal platform (hypervisor, OS and app suite) that can move between different platform types, there would be one for each type e.g. Mobile, Laptop etc?

2) Given that this appears to be platform virtualization rather than application virtualization. If a software house wanted to write an app once and have it be truly portable with this concept, they would need to include JEOS (Just Enough OS)? Given the diversity of hardware platforms in this mobile space, it would seem hard to come up with JEOS to cope with the permutations. Also, in your example of a company moving from an iPhone to a Blackberry, with Platform virtualization, you couldn’t take your app without taking the OS as well which apple probably wouldn’t allow. I know it was probably an off the cuff example, however the example shows off aspects of VMware’s strategy that I’m struggling with.

Paul