One of the great advantages of the absurdly high powered machines we’re all running these days is that virtualisation is more than viable even for performance-hungry applications. Meanwhile Open source continues to crack open the hard nut called proprietary. Two recent pieces of news struck me and I thought worth capturing.
First off IBM acquired Transitive. This is major news. With Transitive’s technology POWER-based systems can run x86 applications. IBM is on a major tear around x86 consolidation- for better management, energy consumption and so on- and this acquisition should significantly accelerate its efforts to sell customers on the notion of data center consolidation.
Over lunch yesterday IBM Software Group general manager Steve Mills said something extremely smart, and absolutely relevant to the deal.
“Tech used to be about displacing labour. Today tech is about displacing tech.”
Of course consolidating servers also potentially allows you to cut staffing levels… which in the current financial environment is a more pressing concern than ever. We used to say Do More With Less. Now we’re going to live it. What’s the end game- Windows servers running on IBM mainframes is one possibility, though I can imagine negotiations between Microsoft and IBM about production support for customers are likely to be rather interesting.
Byron Dennis sees things rather differently, talking of “choking off user choice”.
The other bit of progress that struck me pretty squarely lately as part of the same broad trend, but here driven by open source rather than virtualisation is the Moonlight project, spanning Microsoft Silverlight 2 and Linux with Novell as the intermediary. Here’s Miguel. This is not a totally isolated effort- Microsoft is also working with a French software company to see Silverlight supported through Eclipse. The point here is that there is now developer tool and runtime support for Silverlight in open source environments. Also bear in mind that “the full Silverlight Toolkit is available under the Microsoft Public License, an OSI-Approved license that allows full reuse of the code.”
We’re increasingly living in a world where people say things like:
“The above program was built on MacOS, the result copied to Linux and then executed using the LinuxPlayer. This is still very basic, the port is yet far from done.”
[italics mine]
What effect will Open Silverlight have? Likely more open Flash. As Chris Messina aka @factoryjoe put it yesterday on twitter:
“It would not surprise me one bit if Adobe open sourced Flash to co-opt Silverlight. And to make Flash content web-lucrative with Google.”
Bear in mind that Netflix is now supporting video on demand for Macs using Silverlight, so this is not all BS. The pressure on Adobe is real.
A big question after Microsoft’s recent PDC was whether you could deploy Linux to the Azure cloud. Given the powerful virtualisation technology in Windows Server that’s a business, rather than a technical decision for Microsoft.
Am I saying that proprietary technology is dead? Certainly not – many of the smartest developers on the planet are currently running headlong into the embrace of Apple environment optimisation, for example. But open source and virtualisation are dramatically changing the economics of enterprise computing. Write Once Run Anywhere- we didn’t get there with Java, but old mainframe technology is now changing the game.
Al says:
November 20, 2008 at 4:59 pm
Hi James great conversation starter and interesting post especially around the IBM virtualization of x86 concepts.
But to me virtualization is an interim vendor cloud play “The virtualization rainbow” as I expand on it in my response here:
http://blog.folknology.com/2008/11/20/gold-at-the-end-of-the-virtualization-rainbow/
But we both agree on Open winning more than proprietary moving forward.
regards
Al
Neil Ward-Dutton says:
November 20, 2008 at 7:07 pm
Two observations James:
1) re: Steve Mills – he’s right, when talking about mature markets – but in “growth”/emerging markets, it’s still quite a lot about tech displacing labour. It’ll be interesting to see to what degree IBM manages to reflect this in the way it attacks mature vs growth markets.
2) re: virtualisation, of course VMWare’s announcement of work on a mobile platform is especially delicious in the context of WORA on mobiles. MIDP didn’t cut it as it placed tough constraints, took too many resources, and fit badly with many mobile OSs. Plus there were many other intriguing alternatives as well as native development, all on the table. Virtualisation is an inelegant but deeply pragmatic way around this. Be interesting to see what the power/battery life implications of this approach might be though…
Rich Sharples says:
November 20, 2008 at 7:43 pm
“What’s the end game- Windows servers running on IBM mainframes is one possibility”
As always the devil is in the details. V12n allows server consolidation but expertise consolidation – not so much. Hosting MS Exchange on z/OS or Linux doesn’t mean you can fire your Exchange experts.
Computing ‘machinery’ prices are continually dropping; expertise is a rising cost. We’re already at a point where 1 person can manage vast networks of machines so I don’t think V12n will help reduce $person$ significantly.
– Rich
Aneel says:
November 21, 2008 at 4:42 am
Hey James, since you asked.. geared solely to the Transitive thing. Remember that I’m just a lowly product manager, so take my opinions with a grain of salt.
It’s been said for years that you can replace big (or moderately big) iron with rows of racks of commodity hardware running sophisticated software–everyone can be Google.
1. In fact, very few can be Google. How many companies do you know that are willing to put in the engineering and analytics effort it takes to actually do what Google does well?
2. Server sprawl and underutilization has gotten ridiculous. It’s eating up dollars, space, power, etc. This has turned into a truism. Consolidation + virtualization have helped to some extent, but not nearly enough. And now there’s VM-sprawl eating up utilization and creating management headaches unnecessarily.
So.. now IBM could say: we’ll replace your rows of racks of commodity hardware with a single bit of big iron. It’ll consume less space/power/etc, produce as much or better, have understood and long-tested measurements and benchmarks that you can run automated analytics against, be easier to manage, take less manhours for ongoing maintenance/attention, be backed by IBM support, etc., etc., *and* you can run all your bloody x86 software on it.
🙂 This stuff is cyclical.
Bruce says:
November 21, 2008 at 1:23 pm
Of course consolidating servers also potentially allows you to cut staffing levels… oh really?? No so. at all. In the old days of server attached storage (pre-virtualization) I did not need 2-3 storage architects. Today storage architects are a mainstay of every virtual shop. In the old days I may have had 100 physical servers. Now I have 10-15… but I have 150+ virtual servers; each requiring support. Oh yea… we also need VmWare ESX guru’s. We used to get by with just an MCSE; today we have certifications for VM; Storage; etc. Indeed we are getting way more bang for the buck but my support team has never been more diverse or specialized.
James Governor’s Monkchips » What Should Sun Do? says:
November 26, 2008 at 7:19 pm
[…] – The Storage Is The Computer. The Network used to be the computer, but now its the storage. Everything runs on everything. What matters is how fast you get to the bits and how flexibly you can work with them. That is why […]