Are two OSes better than one?

As Microsoft’s travails in delivering the next version of Windows demonstrate, building an operating system is a bitch. It’s complicated and expensive, and trying to speed up the process often just slows it down. But even as the world’s largest software company struggles to meet its latest deadline for unveiling Vista, it’s announcing a massive new effort to build a completely different operating system – a web operating system modeled on Google’s sophisticated network of computing “clusters.” Google’s Web OS runs the company’s search engine and a rapidly expanding set of related services such as the lucrative AdWords advertising system. Seeing such web services as the future of software, Microsoft feels it has no choice but to build its own Web OS to compete with Google’s. “I believe that only two or three companies can really deliver the infrastructure,” said Microsoft CEO Steve Ballmer last week in announcing Microsoft’s plans.

A Web OS is very different from a traditional computer operating system like Windows. The latter is an intellectual construct, a set of instructions written by people. You have to spend a lot on salaries and other labor costs to build a traditional OS, but the capital expenses are negligible. A Web OS, on the other hand, takes physical as well as intellectual form. If you want to control the OS, as Google and Microsoft do, then you not only have to write the instructions but you have to buy, assemble, and maintain all the equipment – processors, storage drives, and so on – that the instructions control. Because a Web OS runs centrally, you have to own the equipment it runs on, rather than offloading that headache onto your customers.

That requires a lot of investment. Google this year will buy at least $1.5 billion worth of capital equipment, double what it spent last year and about five times what it spent two years ago. In announcing its first quarter financial results a couple of weeks ago, the company said, “We expect that the growth rate in capital expenditures in 2006 will be substantially greater than the revenue growth rate for the year.” (That’s not a state of affairs you’d want to continue indefinitely.) Google is also spending about $600 million on research and development this year, a number that’s rising sharply as well.

Microsoft, said Ballmer last week, will jack up its capital spending for its web business to $500 million during its next fiscal year, which begins on July 1. That’s up from an estimated $300 million this year and just $100 million last year. It will also spend a whopping $1.1 billion on R&D for its web business next year, as it rushes to catch Google.

Clearly, this is a high-stakes battle.

I think its outcome will hinge, to a considerable degree, on whether having two OSes – the computer OS (Windows) and the Web OS (Live) – proves an advantage or a disadvantage for Microsoft. On the one hand, it would seem to be a hindrance. One of Google’s great strengths has been its lack of a past. In creating its Web OS, it’s had a clean slate. It hasn’t had to make compromises with other business units or with existing products. It’s constructed its Web OS’s cluster architecture out of cheap commodity components and a custom-built version of Linux. It’s hard to picture Microsoft giving its engineers a similarly clean slate as they develop a competing cluster architecture. Can you imagine Microsoft announcing, for instance, that it will use a form of Linux to run its Web OS? The existing OS and the business it represents will, in other words, constrain and complicate the development of the Web OS. Microsoft’s web team will have to make compromises. In general, constraints, complications and compromises aren’t good.

On the other hand – and here’s where Google gets nervous – the Web OS is not going to suddenly displace the computer OS. The two OSes will run in tandem for the foreseeable future, with tasks steadily shifting from the computer OS to the Web OS. Microsoft has an opportunity to integrate the two OSes in a way that can put Google at a significant disadvantage. To put it another way, it has an opportunity to manage customers’ transition from the computer OS to the Web OS in a way that furthers its own interests – and damages Google’s. Google could, of course, try to counter Microsoft’s advantage by offering its own computer OS – a version of Linux, for instance – but it’s hard to imagine such a move succeeding. Would a critical mass of users really make the leap?

So, to boil it down, the outcome of the competition between Microsoft and Google may come down to a simple question: Are two OSes better than one? The answer is largely in Microsoft’s hands.

6 thoughts on “Are two OSes better than one?

  1. vinnie mirchandani

    Nick, it is bit more complicated than that. Web OS as you call it works fine for apps that can be parallelized. Many enterprise apps, unless completely rewritten will continue to need UNIX, LINUX, gosh even MVS…OS fragmentation is not going away that soon…

    My view on the grids both are building are they are more of a threat to outsourcers and to some extent the telecoms. They are cannibalizing maintenance, upgrade etc services with the SaaS model.

  2. Dan Ciruli

    Hmm…trackbacks seem to be gone. My comment got a bit too long, so I put it in a post.

    Suffice it to say: if Nick is right about Microsoft’s intentions, and they can bridge “the two OSs,” Google won’t have a choice: they’ll have to release an OS. That should be fun to watch.

    And, Vinnie: I agree that many enterprise apps run on OSs that are neither Microsoft’s nor Google’s, and would need to be rewritten to run on either (and I don’t think that will happen any time soon).

    But I think you’re limiting the scope of what can benefit from some sort of distributed computing. It doesn’t need to be a “parallelizable” app per se: many forms of SOA, for instance, can be run on a grid not for speed’s sake, but for scalability’s sake. In other words, you don’t need to deconstruct your web service so an individual call can run on many machines; rather, you take advantage of a grid so that each web service call runs on an available machine.

    Imagine writing a web service and knowing that, because you are partnered with GOOG or MSFT, your web service can instantly scale to thousands of machines if necessary. It’ll be a powerful tool, indeed.

  3. Tom Foydel

    If Google put together a desktop operating system I would definitely be interested. What value does MS give me? Very little. Bring it on, Google!

  4. Rogel

    completly not connected to the issue: how did you add the pop up layer that jump in when hovering above any Amazon link?

  5. Filip Verhaeghe

    I don’t see why would Microsoft want to switch to Linux, since they don’t have to pay licenses for their own OS. By the way, Microsoft Research actually has a new “clean slate” operating system as well, called “Singularity”, although I don’t think Microsoft will be betting on this for the near future Web OS.

    But the OS underlying the Web OS may not be the real issue. Today, Google may have a great Web OS, but it isn’t sharing that OS. All of the applications running on it are written by Google itself. For all the talk about mashups, real applications as written by indepent software vendors (ISV) cannot be deployed on Google’s “Web OS”.

    Yet part of the reason Microsoft is so dominant, is that so many ISVs are building their products for Microsoft Windows (only). Many of these products are at least in part server products, where Linux is a real option. For these ISVs, the development tools Microsoft provides make the difference. That’s why running around on stage at a Microsoft conference, crying out “developers, developers, developers”, makes a lot of sense. Operating systems (and especially upgrades of the OS) are usually sold through applications that matter to the customer.

    Microsoft being Microsoft, I assume they will try to provide their partners with the best software development tools on the market for Web OS. And that may be where Microsoft get its advantage.

  6. markb in NJ

    Hi Y’all.

    I find it very persuasive (and I almost feel like a google booster or stockholder) when I say this BUT. (I think very seriously that there is a master plan (for world (or at least US) dominination (and Im for it)

    Based on Bob Cringeley(

    old discussions (need to go WAY back), as well as a whole bunch of new things (such as this:

    These continue to lead me to belive the following

    (presuming that some rumors are correct:

    (based on these following legs:

    *google data center in a tractor trailer

    *google dark fiber to many pops around country

    *Google’s TV server sw

    *possible plans to offer TV commercials

    At any rate, it seems [to me at least!] to add up that these all lead to a master plan.

    (which I cover here:

    and Here:

    Bottom line? I’d say the new sql over multiple boxes is a very good tool for the guys to use for their world domination attempt —

    WHAT do YOU think??

Comments are closed.