According to Mark Russinovich, co-author of Microsoft Press' Windows Internals and Inside Windows 2000, Microsoft will preserve its software dominance over alternatives such as Linux for the next two to four years, chiefly because the open-source community doesn't have the same cash and resources to tap up as the world's largest software company.
Windows "still has a much more focused organisation looking at high performance", Russinovich said at this week's Microsoft IT Forum event in Copenhagen.
"Linux improvements are advancing quickly," he said, but added: "Linux efforts are kind of all over the place -- each vendor, like IBM, has its own Linux labs," which don't have the same scope or time to devote to boosting performance benchmarks over time.
He also added that budget constraints are an issue, citing the example of the Open Source Development Labs' $10m budget. In contrast, Microsoft has poured $6bn into research and development this year across all its operations.
That figure may well narrow over time but for open source to raise its game further for enterprise use Russinovich believes the issue of backward compatibility will have to be tackled. "Linus has no problem breaking backward compatibility in device drivers," he said, adding that created a "big problem" for companies today.
System administrators don't want to be recompiling code when a new kernel comes out, Russinovich said.
Jean-Philippe Courtois, CEO of Microsoft EMEA, repeated the company's stance on competition, saying it is good for Microsoft.
"We're serious about Linux," said Courtois. "There's pressure from competition, absolutely [but] competition really provokes our best innovation," he added, claiming the dynamic of discussions on Linux versus Windows has changed.
"It's much less emotional... it's becoming more rational," he said.
Re: R&D budgets of open source software (OSS) vs. Microsoft.
You write: "Open Source Development Labs' $10m budget. In contrast, Microsoft has poured $6bn into research and development"
In OSS the vast majority of contributors do so on their own time -- their effort goes unpaid but that doesn't mean it ought to be unvalued.
For the sake of argument, let's say these efforts are value at $1000m ($1 billion). At $20/h median and 200 hours/annum/developer this would require 250,000 developers world-wide.
This is probably an over-estimate, so let's say 50,000 developers (10m hours) world-wide for a total of $200m at $20/h. Additionally, ten (10) of these volunteers will equal one 40 h/week, 50 week/year formal R&D employee, so this effort is equivalent to 5000 regular R&D employees.
There are incidental expenses that Microsoft's R&D dep't will have to pay (hardware and real-estate) but these are a mere fraction vs. salaries.
This is a wild guesstimate of what's going on in OSS but it does give you a sense of the magnitude of the (not-so) "free" labour (of love) devoted to OSS.
There's also the consideration that ever increasing numbers of programmers are being groomed in China and India (total population > 2000m) and these gov'ts are also likely to devote formal resources to OSS. And, then there are such projects as the various localised Linuxes in Spain, Brazil and elsewhere that have gov't resources devoted to them.
In summary, the R&D budget for OSS is spread out across the world and not easily enumerated but it's significant and growing.