[Info-vax] OT: Arun Kishan

[Neil Rieck]

On Dec 28 2009, 7:09 am, Neil Rieck <n.ri... at sympatico.ca> wrote:
> This is a repost because the previous link required you to login to
> msdn. Sorry about that. (NSR)
>
> This detailed description of Windows Process/Thread dispatcher (old
> and new) seem to have many similarities to OpenVMS. The old was was
> written by Dave Cutler.
>
> Video: 55 minutes
>
> http://channel9.msdn.com/shows/Going+Deep/Arun-Kishan-Farewell-to-the...
>
> p.s. Part way through, Arun mentioned that changes to the dispatcher
> were necessary to support scaling windows onto a larger number of
> CPUs. The last time I heard someone give a similar (scaling) speech,
> it was Gaitan D'Antoni describing changes in OpenVMS-7.3-2 which,
> IIRC, was related to erroneous serialization on IOLOCK8. Likewise,
> both OSs have enjoyed improvements as engineers have analyzed problems
> getting their OSs running in a virtualized environment.
>
> NSR

Lots of people are still sending me emails bashing my Windows posts so
let me set the record straight:

1) I love OpenVMS and consider it my favorite OS

2) Windows is not (yet) as stable as OpenVMS and is still vulnerable
to many kinds of infections

3) After reading lots of books this year including "Showstopper! The
Breakneck Race to Create Windows NT and the Next Generation at
Microsoft" by G. Pascal Zachary I have come to the conclusion that
someday soon (perhaps 5 years) Windows will be as stable as OpenVMS.
Why? Because of the influence of the ex-DECies at Microsoft.

4) The ex-DECies at Microsoft have made Windows-NT (and 2000, XT,
Vista, Win-7) more respectable. The code bases known as MS Chicago and
Cairo have been totally replaced with ex-DECie code bases like
Whistler etc. and it shows. Too bad MS stuck with the name "Windows"
because everyone associates the new stuff with the old stuff.

5) I have done some fancy OpenVMS I/O programming where you set up
conditions for communicating with the kernel and can tell you that
this Microsoft video:

http://channel9.msdn.com/shows/Going+Deep/Arun-Kishan-Farewell-to-the-Windows-Kernel-Dispatcher-Lock/

which indicated where Windows came from kernel-wise (Cutler and VMS)
and where they are going (re: 256 core support) reminded me of the
story of the "Tortoise and the Hare". We OpenVMS hares are satisfied
to sleep with point releases of OpenVMS every 18 months while the
Windows tortoises are catching up fast. The execs at HP see this too
and are just biding their time while providing minimal life-support to
OpenVMS. They make hardware which runs multiple operating systems so
really don't give a damn.

6) Just as we have all seen with Solaris and OS-X when they've been
installed on quality hardware, Windows will also run better. Sure
there was the big fiasco a couple years back at the London Stock
Exchange but sometime in the future someone is going to try a stunt
like that again and it will work. When that happens programmers who
can only write code for OpenVMS will be put out to pasture.

Neil Rieck
Kitchener / Waterloo / Cambridge,
Ontario, Canada.
http://www3.sympatico.ca/n.rieck/