[Info-vax] New CEO of VMS Software

[chrisq]

On 1/7/24 19:47, Dan Cross wrote:
> In article <memo.20240107190811.16260s at jgd.cix.co.uk>,
> John Dallman <jgd at cix.co.uk> wrote:
>> In article <unehno$14f82$1 at dont-email.me>, arne at vajhoej.dk (Arne Vajhøj)
>> wrote:
>>
>>> Sun had a problem:
>>> - Solaris/SPARC servers were more expensive than Linux/x86-64
>>>    servers
>>> - the applications running on Solaris/SPARC were typically not that
>>>    difficult to port to Linux
>>>
>>> Asking for a premium without sufficient vendor lock-in is a bad
>>> business case.
>>
>> Their responses were also not that great:
>>
>> They open-sourced their OS in the belief that this would "reduce
>> development costs" as Linux people switched to working on Solaris. This
>> didn't happen to any noticeable extent. The open-sourcing part created
>> lots of work for expensive lawyers and slowed software development.
> 
> I know many of the players involved with the creation of
> OpenSolaris and I think they would dispute parts of this.
> 
> It is true that the hoped-for shift of open source people
> working on Linux and BSD moving to working on OpenSolaris did
> not materialize.  But why?
> 
> I believe that, within OpenSolaris, the feeling was that Solaris
> was "so obviously better" that people would just naturally
> gravitate to the technically superior offering.  However, by
> this time, Linux was "good enough" and improving rapidly;
> certainly at a pace greater than Solaris was improving.  So
> while there were parts of Solaris that were (and arguably still
> are) technically superior to Linux, the feeling was that Linux
> would overtake Sun in these areas soon anyway, so why switch?
> Secondly, a lot of people were put off by the CDDL; Linux seemed
> safer and more "free."  Moreover, some parts of the operating
> system remained closed, and you pretty much had to use SunPro
> (at last at the beginning) to build things, and that was still
> proprietary.
> 
> That said, while the initial open-sourcing was expensive, it is
> not clear to me that the ongoing cost was particularly high.
> Certainly, I have _never_ heard anyone who worked on it complain
> about the ongoing cost.  Re-closing the source code was highly
> contentious.
> 
> I think the reason OpenSolaris failed was that it was just too
> little, too late.  There wasn't a good reason for people to
> switch.
> 
>> Cut back their hardware development, since it was expensive, making their
>> systems even less competitive.
> 
> Yes.  They really missed the boat on x86.
> 
>> They ended up selling themselves to Oracle, of course. Oracle's plan was
>> vertical integration: tuning up SPARC and Solaris hardware for Oracle
>> database so they had a price-performance advantage on their own hardware.
>> A great plan, except that the tuning had already been done and there was
>> no unrealised performance available.
> 
> Well, when the main reason your systems are sold is to run one
> program specifically....
> 
> It's a shame.
> 
> 	- Dan C.

Back in the day, Vax etc, software was optimised and fine tuned
to match the hw is ran on, so perhaps the Oracle "engineered
systems" idea was just updating that concept. If you look at
the last Sparc release docs, theres's a lot of database and
high speed comms related included in hw. Far too expensive,
of course, and perhaps the last gasp of proprietary hardware and
os's, which can never hope to match the resources available to
the open source movement.

To be clear, the Solaris sold by Oracle is not the same as Open
Solaris, which was independently developed from the original Sun
source release. Openindiana, in constant development
and a free alternative to the Oracle offering. Also used as the
core of Joyent Smartos and other systems.
Solaris 10 was a major milestone, with the introduction of the
ZFS filesystem, and lightweight virtualisation via Zones, or
containers, whatever they are called now. This was a decade or
more ago. The FreeBSD clean room ZFS implementation eventually
became OpenZFS.

Finally settled on FreeBSD partly because that too had ZFS, a
similar lightweight virtualisation implementation, a very
disciplined development process and more. No systemd either.
The clean room ZFS implementation eventually becoming OpenZFS.
All in all, a worthy successor to Solaris...

Chris