% fortune -ae paul murphy

A quick comment on Singularity

I've received a few emails over the last few days from people wanting to know whether I think that Microsoft's Singularity OS will eventually push Unix aside, first in academia and then in business.


Singularity has an odd history - a research project that became an executive agenda item in 2004/5 and has since been relegated to research status once again.

The executive dream, circa 2004, was first that Microsoft could develop a non Windows network OS that could be leveraged to flip the company from Xeon to Xenon and secondly that one of the then in progress internal OS development projects would provide the basis for it.

I believe that the first part of this is unrealistic because Microsoft's internal structures work against innovation - and that what happened with Singularity pretty much illustrates that.

That project started as a vehicle for research aimed at working through the consequences of trying to combine a Minix style kernel just barely capable of running processes in simple device spaces with OS/400 style code verification on x86 hardware. To make that work without first having to re-invent the wheel they used parts of the dot net technologies to substitute for the near hardware level microcode IBM uses - and that, in turn, required them to rely on a lot of innnovation defeating Microsoft standard technologies like the executable file framework.

There's a joke about there being nothing more dangerous to researchers than high level attention and what happened next nicely illustrates that: expectations clashed, fiats were issued, committees met, compromises were made; spin-offs demonstrations - including a Xenon version that apparently ran like an underwater tricycle - and meanwhile the surviving members the original team tried to keep their heads down and carry on and a few higher ups tried to butress their positions through judicious information leaks to the press. Out that process we eventually got what we have today: a little experimental system built on the Windows foundation and roughly on par with what you'd expect for a masters level group project at a middle tier school - surprisingly enough largely what it was intended to be, but not remotely a distributable, business class, network OS.

So is Singularity a threat to Unix? No, it's not even a threat to Windows - all of the hot air being blown around on this ultimately originates in the scramble for cover consequent to the clash between uninformed executive expectations and the thing's real purpose.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.