VirtualBox 3.0 Beta: Finally something exciting in desktop virtualization

by with No Comment computersITTechnology

 

Sure, but can it run Excel? –
The 3d animator in me appreciates big iron, massive compute clusters and grid computing.  As my compatriots at Pixar and LucasFilm will attest, there is NEVER enough computing power for what we want to do.

But my day job is to handle the computing needs of a top tier Business School where the research is much less dependent on raw computation as it is dealing with large datasets and iterative processes.  Cluster computing is the new hottness in academia, but for the uninitiated Windows user, cluster computing is quite limiting, as one’s compute jobs must be written in such a way as to make it distributable, and that’s not usually as simple as hitting “Go”.  There is still a world of research out there for which clusters are either not useful, or are still too much of a hassle to bother with.  There bulk of the research world can still benefit from having familiar desktop tools shared on a single large computation resource.

So then, what’s the answer?  Terminal Services is a real pain to manage and application compatibility is problematic.  Virtual Applications would be nice, but we have yet to see a Virtual applications platform that handles 64-bit computing, and again, application compatibility is problematic. 

Load em up, Roll em out, then put em out of their misery –
So my big idea was to provide “on-demand higher-performance” virtual machines.  VM’s are quick to provision and dispose of, without permission worries and other “administrative overhead”, so it just made sense to give a user temporary access to a VM with a limited software set, let them run their computations using the interfaces they already know… and then delete the VM.

I have been trying to do something similar myself.  Many of the 3d applications I use come with free licenses for distributed processing of the final frames of animation.  Cluster computing may be new to the general computing world, but tell that to the 150 node rendering farm I set up in ’94.  The problem has always been in the setup and maintenance of the software, not so much in the ability of the software to distribute computing load.  Instead of installing the software a hundred times, it would be far better to configure a VM once and distribute that to each host. I have had some success with this method under ESX and Hyper-V, but they aren’t exactly "desktop" solutions, and they only scale so far. Without being able to access all available processing power of a physical host through the VM, neither vendor offers a real solution for this purpose. 

That has remained on the wishlist for years.  Straight up Desktop Virtual Machine technology has crept along very slowly, as vendors have rushed other kinds of virtualization products to other markets and the major vendors have spent their time fighting over server virtualization.

So many technologies, so little polish –
The virtualization market is flooded with potentially excellent products that could be used in so many more ways… if only they did "x", or if only they supported "y" without crashing.  While a reasonably priced computation server could have as many as 32-logical CPUs today, Hyper-V only allows a VM to use 4 of those, and VMWare’s newest version of ESX now brings only 8 to the party.

I have tested both solutions, and both utilize their virtual processors as well as physical hardware (within a couple percent), so the problem is one of scale, not capability.

On the desktop it’s worse.  VMWare Workstation has long been the only option for multi-core, 64-bit VM’s, and it only supports 2 virtual processors.  VirtualPC is way, way, way behind the technology curve and doesn’t support multiprocessing at all, much less graphics acceleration, even with Windows 7’s new version.  And neither solution accepts its competitor’s virtual hard drive format, which makes both solutions less attractive.

As an aside – I really think Microsoft blew a huge opportunity to sieze market share from VMWare by failing to produce a desktop variant of Hyper-V.  EPIC FAIL!!  The latest VirtualPC is years behind every other offering, and I simply could not look a client in the face and tell them to use "XP Mode" as a way of maintaining compatibility with their XP applications.  Someone really should get fired for that… but I digress.

Sun’s VirtualBox project has been the lesser known Swiss Army knife of Virtualization, accepting VMWare’s and Microsoft’s virtual hard drive formats in addition to its own.  VirtualBox provides most of the functionality of VMWare Workstation, and more, for FREE… but it has always had one major failing in my eyes: No Virtual Multiprocessing.

But then the silent dragon made a puff –
All that is about to change.  VirtualBox 3.0 beta1 is available… and in addition 64-bit VM support, experimental DirectX support and “Seamless Mode”… it supports NOT 1… Not 2… not even 8… but 32… yes, 32 virtual processors.  As I said, that’s a quad-quad-Nehalem’s number of Virtual Processors.

Sadly, it is still beta.  On my Core i7 machine, running Server 2008, I was able to configure a virtual machine with 8 virtual cores, but it would only launch with 4.  There is debate as to whether VirtualBox recognizes hyperthreads as virtual cores; I am testing this.  Windows 7 setup will not even run without certain options enabled (you’ll have to guess).  The graphics acceleration seems to only available in 32-bit WIndows guests at the moment, so no Aero in x64 yet.  Beta difficulties aside, I did manage to get a new machine installed on a .VHD and managed to import that machine to Hyper-V.  I was also able to take a VMware Workstation machine into VirtualBox, though there were some odd crashes with that.

Nevertheless, this is a good thing and it’s nice to see a significant competitor bully its way into the stagnant realm of Desktop Virtual Machines.  Being able to virtualize without the traditional penalties of virtualization will open many doors beyond simple server consolidation.  My “on-demand” model for high"er" performance Windows machines is likely to become reality.  I would like to see our laboratory and kiosk machines running as virtual appliances, unbeknownst to the users… portable highly secure administrative machines… checkout VM’s for using high end or temperamental software in a sandbox.  Which precise combination of virtual machines, virtual applications, presentation virtualization and so forth wind up saving the most time and money is yet to be determined, but clearly we’re finally entering into the days where the performance differences between the virtual and physical worlds are coming well within acceptable ranges.

If you’ve got the stomach for beta, give VirtualBox 3 beta 1 a try.  I like it already… can’t wait for the final.

j.ello on IT Part 1: The New Old Thing

by with No Comment computersITj.ello on ITTechnologyUsers

I have been in the so-called “IT” profession for more than half my life.  I am shocked to find how little has changed in that time… and how much everyone thinks it has.

I was pleased to find a lovingly-written article on the perennially reborn, long-beloved operating system called GEOS.  If you haven’t heard of it, it provided a Mac/Windows-like interface on a Commodore 64/128… in 1986.  Think about that for a moment.  There was a mouse and a friendly, visual operating system for a computer that couldn’t even hold a candle to an average cell phone.  In fact, the “Start Menu”, so synonymous with Windows, was a concept introduced years earlier in GEOS. A WYSIWYG word processor, a spreadsheet application, a database, and a whole lot more.  All the fundamentals of a full fledged office suite, and while no, GEOS wasn’t the first to do these things, GEOS did them all in 64k, 20 years ago.  That says something about us, about technology, and about the Information Technology industry.

Back to 2006, it seems we’ve come so far.  But after 20 some-odd years of windowed-interface development, what are the primary activities of a typical person at work in front of a computer?  Office applications. CPU speed, RAM, storage, all multiplied thousands of times over 20 years, and we do office applications.

“…why, oh why, do people still work at the same speed no matter how fast their computers are?”

And you’ll say, “but we do so much in addition to that”.  Do we?  I could send email in 1986.  We had the equivalent of your typical websites, and while they weren’t all interconnected or particularly pretty, they did have daily content, forums, some even had multi-user chat rooms.  We had downloads such as midi-like renditions of music (illegally created and distributed, just like today), software (freeware, shareware and pirated, just like today), and such teen favorites as porn and the “Anarchist Cookbook“.

Truth is that the desires of casual computing haven’t changed, neither has the amount of knowledge necessary to do the things we want to do.  Entertainment *STILL* almost single-handedly drives the economics of computing, with no end in sight.  All we’ve been doing for 20 years is applying new capabilities to make the old ideas more attractive, more intelligent, more “2.0”. 2.0… oh yes… I’ll come back to this “2.0” crap… just you wait.

So, if all we’re doing is the same old thing, why isn’t it any easier?  Why are our IT staffs inundated with seemingly basic questions… better yet, why do we still have IT?  Many technology analysts have been predicting for at least as long as I’ve been alive that all of this will simply become black-box and push-button, akin to electricity.  And yet here we are, with an ever diverging technology base, and an ever more complicated set of scenarios for the IT staff… and an ever developing understanding of how inadequate our infrastructure is to handle the “same old things” that we’ve always done.  Why do we insource, then outsource, then insource again?  And why, oh why, do people still work at the same speed no matter how fast their computers are?

These answers and more will be revealed over the next few days, including why all the experts preaching “computing is commodity” are wrong… and why I say “Computing is art.”

j.ello on Tech Op/Ed

by with No Comment computersITj.ello

Well hello world, glad to see you again.  If you don’t recall, seven years ago I started a little e-zine called the j.ello | report.  I kept up with it for a couple years, had a good number of subscribers… but, as it was a labor of love as opposed to revenue based business, I had to put the writing aside for some other projects.  The last 5 years have been quite noteworthy; I’ve been remiss to let them pass without a peep.  Nevertheless, I think even more interesting times are in the near future, especially with businesses, politicians, lawmakers and judges flailing blindly about the place trying to get some kind of handle on the evolving Internet economy. 

“…the human equivalent of a Head-On Commercial.”

So what’s up with this Tech Op/Ed “blog”?  This site is for articles more focused on the evolution of Technology, through business, law, and notable events.  Who doesn’t have a few choice words for Ted Stevens’ “series of tubes” quote?  It’s perhaps *the* most ironically ignorant commentary on the nature of the Internet of all time.  Who’s Ted Stevens?  The chairman of the committee that regulates the Internet… and as far as technology goes, the human equivalent of a Head-On commercial. 

But I can’t be too down on Ted.  In fact, he may not even be an idiot.  He’s actually quite representative of the level of understanding most people have on technology topics.  The problem is that while such massive ignorance exists within entities that directly control the dissemination and regulation of technologies, we will continually find ourselves at the prickly end of the legal and regulatory absurdities that have defined recent times in technology.  It’s a problem the industry faces both externally and internally.  Despite the *vast* amount of “expert” pontification out there, I am still hard-pressed to find an individual who can define even a ubiquitous term like “Information Technology”… and even fewer organizations that have a handle on it.  We’ll come back to this. 

That’s what Tech Op/Ed is about, but what’s in store for the j.ello | report?  One of the reasons I stopped writing the j.ello | report is that it took longer to maintain the site than it did to actually produce the content.  The site would have been (and will be) greatly improved by the more mature “blogging” tools that are available today.  So by experimenting with Tech Op/Ed, I’ll be able to resurrect the j.ello | report in a more manageable form.

In the mean time, I’d like to produce some articles that make a point and are worth reading.  And with that… on to the first post.