[Info-vax] Should Oracle buy OpenVMS?

[ChrisQ]

Bob Koehler wrote:
> In article <0036b65c$0$30964$c3e8da3 at news.astraweb.com>, JF Mezei <jfmezei.spamnot at vaxination.ca> writes:
>> When Sun changed the pricing to Solaris (to $0), did that greatly reduce
>> the TCO of Solaris/Sparc solutions, or did it just shift costs to
>> support and TCO remains 10 times higher than for Linux ?
> 
>    Initial cost of ownership for Sun is still higher than Linux.  A lot
>    of folks are in charge of project development and will look only at
>    development cost and not TCO.

Please elaborate - it may have been true in the days when Sun charged 
for Solaris, but now that it's open source, the only thing remaining is 
the support cost. I would think that is fairly similar, as it's 
primarily labour intensive. Both install fairly easily and at least 
solaris recognises the video card transparently. More than can be said 
for Suse 10.? which I recently had a look at. Came up after install at 
vga 640x480 and needed quite a bit of fiddling to get a sane resolution.

It's stuff like that gives linux the slightly hacker feel, to which you 
can add the over blown and un neccessary desktop decoration and other 
gimmicks, which you have to (again) spend time turning off. Solaris just 
comes up with cde, whatever and you never have to frig about with it 
again. I want to *use* an os to do real work and not have to delve into 
the depths to make stuff work, other than out of interest.

> 
>    This may not be good strategic planning, but it is real.  And I've
>    seen no string evidence that Sun has a lower TCO than Linux these
>    days.
>

An old friend of mine recently told me that they paid red hat over 750k 
uk pounds (~$10e6) for support last year. Mind you, they do have over 
200 Tbytes online over dozens of dl380 etc servers. On balance, I think 
the reason linux became popular was because it was free to use, a unix 
lookalike that ran Apache. Critically, it's reputation had matured to 
the point that it was seen as being robust enough and the install 
process became automated and simple enough for medium level competence 
admins to do the work. What everyone forgets, of course, is that linux 
is just as complex as any other unix under the hood, so when something 
breaks, it needs the same level of competence to fix it.

What linux doesn't have, is the holistic (if I can use such a word :-) 
oneness of an os that has been developed under the gaze of the same team 
ethos for decades, ie: vms as it was, solaris, irix, etc. Os's that used 
to be described as industrial strength, designed from the ground up from 
software engineering principles so that all the bits fit together with 
no gaps. If bits come from everywhere, how do you test the whole and who 
handles and controls regression testing when changes are made ?. All 
that can be said about linux is that it doesn't seem to fall over, but 
where is the data to support a given confidence level ?.

One point that no one has touched on thus far is the benefit of running 
on non x86 hardware. While it would only be part of an overall security 
strategy, the fact that intel binaries won't run on the hardware could 
be a significant benefit in terms of virus resistance. Just one more 
reason why the diminishing pool of architectures is a bad thing.

To get back on topic, I can't think of any reason at all why oracle 
would want to buy vms. Hardly any modern software runs on it anymore, no 
one knows how to admin it and it only runs on one current architecture. 
As for security, the rest of the world has probably caught up.

I think they got a bargain...

Regards,

Chris