On Sat, Apr 9, 2011 at 6:57 PM, Ian Collins <i...@ianshome.com> wrote:
> On 04/10/11 09:25 AM, Garrett D'Amore wrote:
>> Right. And in the real world, customers are generally not involved with
>> architectural discussions of products. Their input is collected and
>> feed into the process, but they don't get to sit at the whiteboard with
>> developers as the work on the designs.
>>
> But they are involved in the discussions around which features should be
> there, and help to prioritise those features.
>
> I guess my fear is the external ZFS developers have adopted the Oracle
> rather than the OpenSolaris development model. We all know where that
> leads....
>
> I hope my fears are misplaced.
My *fear* is that ZFS *does* go the way OpenSolaris did...
Two of the biggest strengths of Solaris for me (I have been
working with Solaris since 1995) have been the reliability and the
scalability.
Part of the reliability came from a conservative approach to new
features, yes it meant that Linux might have a new feature before
Solaris, but my Solaris systems were more reliable. Great for a
server, not so good for a Workstation. So I run Solaris on servers and
Linux on my desktop.
The second part (scalability) is illustrated by my experience with
response time vs. system load (the technical definition). On Solaris
systems response time remains good until the load gets to at least two
times the number of CPUs (cores) and generally closer to three to four
times, while my experience with Linux is that once you get to a load
value of twice the number of CPUs (cores) the system has already
become unresponsive. I have run the load on a test T2000 (32
CPU/core/thread) to over 100 and had good response time. When the load
got to 1000+ it got unresponsive. When the load on my core2/duo Linux
desktop gets to two I might as well go get a cup of coffee.
I saw the OpenSolaris project heading in the direction of Linux
(lots of new features, not so much about robustness and scalability)
that I feared what the future held.
I can't afford to have ZFS be anything less than completely
bulletproof, either in my day job or for my personal data. So far it
has been (neither at work or at home have I suffered any data loss,
even with a myriad of drive failures and failure modes). If there is
not *some* control over development direction, then I fear I will
start seeing more commonly occurring bugs that do lead to data loss.
It's not that I don't trust the Open Source Community, I just don't
see the entire Open Source Community as having the same goals (values
?) as what *I* need (and what ZFS has so far met).
DISCLAIMER: I am NOT a developer (just look at any of my shell scripts
or PHP code) but a System Administrator, and I *know* it and leave
development to those with that skill set. I DO provide input in terms
of features whenever I feel I have something substantial to add. And,
YES, I have had simple feature requests REJECTED by high profile Open
Source projects such as SAMBA because I needed a feature that "no one
else could possibly want". So we modified the source code and ran our
own spin of SAMBA (and hated doing it for a whole host of reasons).
Ping me offlist if you really want to know what feature we needed (to
turn off). I do NOT want to contribute code to ZFS, but I would like
to have some input into future features. I hope this new ruling cabal
(not meant as a negative term in this context) makes allowances for
this.
--
{--------1---------2---------3---------4---------5---------6---------7---------}
Paul Kraus
-> Senior Systems Architect, Garnet River ( http://www.garnetriver.com/ )
-> Sound Coordinator, Schenectady Light Opera Company (
http://www.sloctheater.org/ )
-> Technical Advisor, RPI Players
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
