But other firms are pitching for this space too. http://www.ok-labs.com/ Open Kernel Labs (OK Labs) and server virtualization stalwart Citrix (currently a rumored acquisition target for Cisco) said this month that they will collaborate on the Citrix Receiver virtualization client, based on the OKL4 mobile hypervisor.
This combined product will let Android, Symbian or Windows Mobile phones display secure, virtualized desktop images – not as broad in concept as VMWare‘s MVP, but designed to provide easily deployed and securely managed access to enterprise and desktop applications from wireless devices.
OKL last week announced OK:Symbian, a virtualized version of Symbian OS, which can then be used as a guest operating system on top of OKL4.
This lets Symbian developers or vendors run multiple mobile OSes or multiple instances of the Symbian platform on a single device, and migrate Symbian apps to new hardware. Suppliers can also put Symbian on far cheaper hardware to spread the software platform to budget markets, said OKL – another key objective of mobile virtualization specialists.
"This is the first step of the coming together of mobile phone and enterprise virtualization," OKL’s CEO Steve Subar told Linuxdevices.com. "This solution can easily deliver applications to mobile users, including delivering Windows applications on phones that don't run Windows. Corporate users only want to carry one mobile device, so there needs to be an easy interface to open and manage content.”
It showed a version of its VLX virtualization stack for Android, running on an ST-Ericsson ARM9-based reference design. This used a version of the existing VLX-MH (Mobile Handsets), and the firm also makes VLX-DM (Digital Media) and VLX-NI (Network Infrastructure), plus a new variant for mobile internet devices.
Unlike enterprise virtualization products such as VMWare, which abstract all physical hardware for easier installation and configuration, VLX lets guest OSes access physical hardware subsystems directly, for faster performance, VirtualLogix explains. However, developers can securely isolate critical applications.
Copyright © 2009, Wireless Watch
Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.
Is this 100% pure BS or what?
Put the phone to one side for a moment, think of the PC equivalent. A Windows 98 "Personal Information Manager" in a VM does *not* instantly share info with a Windows 7 "Personal Information Manager" in a different VM, even though they are both running "at the same time" on the same hardware. They need common file formats and synchronized access to shared storage, otherwise why bother?
Or is there something magick going on here which I have failed to understand?
For specialised processing requirements maybe.
"The results would be returned across the internet to the phone, speeding up tasks like graphics processing and supporting high end video or gaming. Intel even says CloneCloud would be able to decide dynamically whether a task would be better processed by the device itself or in the cloud, depending on its processing burden and the quality of the network connection."
Currently only somewhat specialised processing requirements benefit from this approach, where the latency requirement isn't fast and bandwidth requirements of the job input/output is small, but the compute/memory demand of the job is high. I do this currently for spam content analysis where the MTA virtual server processor in a datacentre that looks at an email and accepts or rejects it offloads the content analysis to a faster CPU at home with more memory but slower bandwidth. But I don't see either of network bandwidth or latency being fast enough very soon to enable this to be done for graphics processing or high end video gaming, where a lot of CPU and memory has to be very close to the display right in front of the end users' eyeballs. Getting a faster computed response to a deep strategy game, e.g. go, using a Monte-Carlo simulated annealing approach maybe, assuming a very large parallel supercomputer is available over the network link rather than in front of the player. But again, this is a more specialised requirement.
Automating this would usefully require an extra layer in the exec() layer of an application which collects statistics based on input/output and CPU/memory loadings. The administrator of the system in question would need to choose to add the overhead of checking for suitability for selected candidate jobs, or the overhead added by the extra layer involved in checking for suitability of all jobs would outweigh the benefit for the few jobs which would benefit.
Sounds like the solution to a problem
Wouldn't the time be better invested in making the phone's OS better? I fail to see how any of the stuff mentioned in this article can't be done just as well without virtualization.
And storing a complete image of everything on your phone in the cloud? Google must be salivating.