[gentoo-dev] production vs enterprise and the direction of Gentoo

[gentoo-dev] production vs enterprise and the direction of Gentoo

[kashani]

	I really enjoyed last weeks discussion about enterprise, devs, and 
users. But I think a significant number of users don't fall into any of 
those categories... well at least I don't.

	I'm an admin and have only worked in ISP, NSP, hosting, and startup 
gigs. These systems tend to be fast moving, customer and feature driven, 
and a bit on the bleeding edge. I don't believe any of those 
environments fall into what is normally considered "enterprise", but 
they are production systems with a premium on flexibility over stability 
much like a Linux distribution we all know and love. :)

	Running Gentoo in production has been a competitive advantage for us. 
Need tomcat installed? Here it is. Need PHP5? Here it is. Need XML 
nonsense in PHP? Here it is. Need a Mysql driven virtual mail system? 
Here it is. At the time to do the above on most stable or enterprise 
systems would have involved waiting for the packages I wanted to be 
releases 12-18 months after the fact or becoming a dev. Neither are a 
great use of my time. I'll take a few thousand random Gentoo users 
posting on the forums over a QA department of just me. So rather than 
see Gentoo think about doing or not doing enterprise type stuff like 
long releases, support contracts, etc I think most of us LAMP, MRTG, 
Nagios, BIND, Postfix, Qmail, Postgres, thttpd, Courier, etc geeks would 
like "just a bit more stability" rather than the overhead that comes 
with enterprise features.

	Some people really like the idea of snapshots. I'm not completely sold 
on it, but would find it interesting if it were 6 month snapshots or 
less. I like the idea of meta builds. I'd be even more interested in 
snapshotting only the meta build rather than the whole OS. The mail 
server stays stable, but the rest updates as normal... I'm sure that 
idea will turn out to have significant drawbacks. Reverse dependency 
checking... been discussed to death and we all know it's coming. Once 
it's working well that's going to eliminate a number of problems 
especially if it'll warn you at the emerge -upv level that some updates 
are going to require a revdep-rebuild rather than after the fact.

	To be honest /etc/portage/* tends be enough control for me, but I've 
been doing admin work for 9 years. And I've got a test environment that 
I use religiously. I do however answer a lot of questions in the Network 
and Security forum where up and coming admins aren't so lucky. I don't 
think any of us expect a foolproof system, but there is room for a bit 
more stability that'll benefit all users... especially admins. :)

Hopefully this sparks some more interesting debate.

kashani
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] production vs enterprise and the direction of Gentoo

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 2731 bytes --]

On Thu, 2005-06-16 at 12:55 -0500, kashani wrote:
> 	I'm an admin and have only worked in ISP, NSP, hosting, and startup 
> gigs. These systems tend to be fast moving, customer and feature driven, 
> and a bit on the bleeding edge. I don't believe any of those 
> environments fall into what is normally considered "enterprise", but 
> they are production systems with a premium on flexibility over stability 
> much like a Linux distribution we all know and love. :)

Production != Enterprise

I know what you mean, though.  However, you are definitely not the
target for any of Gentoo's "Enterprise" efforts, as they are, by design,
much slower moving and much less flexible.

> posting on the forums over a QA department of just me. So rather than 
> see Gentoo think about doing or not doing enterprise type stuff like 
> long releases, support contracts, etc I think most of us LAMP, MRTG, 
> Nagios, BIND, Postfix, Qmail, Postgres, thttpd, Courier, etc geeks would 
> like "just a bit more stability" rather than the overhead that comes 
> with enterprise features.

Honestly, the more bleeding edge it is, the less quality that is going
to be associated with it, simply due to new features and other such
things that will not have gotten as much testing/bug-fixing as the more
stable features.  Unless we have our own audit team specifically looking
for bugs and fixing them *before they are reported* then there isn't
much we can do to resolve this situation.

> 	To be honest /etc/portage/* tends be enough control for me, but I've 
> been doing admin work for 9 years. And I've got a test environment that 
> I use religiously. I do however answer a lot of questions in the Network 
> and Security forum where up and coming admins aren't so lucky. I don't 
> think any of us expect a foolproof system, but there is room for a bit 
> more stability that'll benefit all users... especially admins. :)

That is one of the problems we have here at Gentoo.  We provide ways of
doing things that you want to do.  Sometimes the methods are very
simple, other times, we expect you to know what you're doing.  The real
problem we have is how to balance between the "do everything for the
user" approach of some distributions and the "don't do a damn thing and
then let the user RTFM" approach that is a bit more common with Gentoo.
Personally, I think we simply need to make sure things get documented
well and leave it up to the community to provide the "support", as it
were.  This has worked quite well for Gentoo, and tends to tighten the
bond between users and developers.

-- 
Chris Gianelloni
Release Engineering - Strategic Lead/QA Manager
Games - Developer
Gentoo Linux

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]