Enterprise Acceptance

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

Will Gentoo be accepted as a supportable distribution in the enterprise?

i.e. Not currently supported by EMC, Veritas, Oracle etc.

Explain your vote.

Don't flame me. :-)

aidanjt · Posted: Sun Aug 05, 2007 12:07 pm Post subject:

I answered 'Yes', but not because I think it will in portage's current form, but because 'ever' would be an absolute, portage could transform to be an enterprise class package manager, so I couldn't answer 'No' because your question was ill-defined.
_________________

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

Thanks. I'm not convinced it is ill defined.

IMHO having a 'moving target' distribution where there is an almost infinite way of configuring or fine tuning a system makes supporting it tricky. Don't get me wrong, Gentoo is great.

Out of interest, what features would turn Portage into an enterprise package manager?

PaulBredbury · Watchman Joined: 14 Jul 2005 Posts: 7310

The question doesn't have much meaning. There's nothing to stop anyone who knows the root password (or has physical access to the machine) from "tweaking" any distro so that it makes e.g. Oracle unstable. Just change one of the libs that Oracle depends on.

Gentoo is one of the most unlikely distros to get "certified" for anything, because of our notorious ricer nature.

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

Ok perhaps I wasn't clear, apologies.

Say for example I wanted to use Gentoo at work and connect it up to our SAN. I couldn't use any EMC product (Powerpath, Networker, SYMCLI) and expect to get support, some just would not work without dicking with the packages. Likewise with VxVM/VxFS even If I could get it to install on a server it would not get supported.

By support I mean help setting it up, or support in fixing a bug (which with EMC software it likely!)

Most enterprises see Binary RPM based distros as the way forward. I see Gentoo as great for HPC, Power Computing or perhaps as a base. Gentoo is good as a server distro until you want to interact with a vendors app.

NeddySeagoon · Posted: Sun Aug 05, 2007 1:17 pm Post subject:

sanman,

I voted yes, even though I think it is unlikely - its still possible.
Gentoo would need to introduce another ACCEPT_KEYWORDS and have it correspond to a defined set of packages at fixed versions.
Both stable and testing are moving targets.
I don't see it happening any time soon because its not what development or Gentoo is about.

Gentoo is a meta-distribution. It allows you to build your own distro the way you like. Thats the exact opposite of whats needed in an Enterprise, where stability and certification is valued above all else.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

Neddy,

Thanks for your comments. I have used Gentoo for a bit, perhaps I had lost sight of the fundamental belief behind it.

Does anyone feel that by not having an 'Enterprise' software branch the growth of the distribution will suffer, and Gentoo will remain as a power-user/hobbyist distro (not that there is anything wrong with that)?

Cheers.

Sanman

PaulBredbury · Watchman Joined: 14 Jul 2005 Posts: 7310

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

jabol · Apprentice Joined: 05 Oct 2005 Posts: 269

Well, I think it is supportable and very well so. It seems to me that major vendors are just to lazy to have a set of packages for each distribution out there...

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

aidanjt · Posted: Sun Aug 05, 2007 3:25 pm Post subject:

Tell the vendors to play ball with opensource, I don't see why we should bend to their will and place limitations on choice. Of course there's always the choice of telling them where to stick their over priced licenses and use a free/open version of what they're offering. They can still protect their IP with whatever license restrictions they want, plus various laws, throwing out binary-only programs on UNIX causes no end of problems.

As for Debian being figurehead for a stable server OS... aghm.
_________________

PaulBredbury · Watchman Joined: 14 Jul 2005 Posts: 7310

"aghm" is not a sensible counter-argument. Suggest an alternative distro, or an alternative explanation as to why the Gentoo Server Project got nowhere.

steveL · Posted: Mon Aug 06, 2007 11:10 am Post subject:

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

I didn't even know about the project until I created this poll. I'm up for the fozen tree/server project. It makes alot of sense to me. I can spare a couple of hours a week so I'd be glad to help.

I appreciate what you are saying, and In an ideal world yes big corporates should open up their software. But in reality very few of the big corps. are going to do it! I don't see EMC opening up their software solutions stack anytime soon - they make far too much $$$ from it. The enterprise is slowy embracing Linux both at the desktop and at the server level, but I don't see many going down the full OSS route. I've worked for three big enterprises now and It is always a mixed environment. Companies like the security of having support contracts etc.

I wish the people I support were all tech savvy. :-)

steveL · Posted: Tue Aug 07, 2007 3:15 am Post subject:

WTF is EMC anyhow?!

enqack · n00b Joined: 16 Feb 2007 Posts: 4

Well I don't see it being so any time soon I do believe it possible and to my surprise I did talk to a person that works within a subdivision of the University of Michigan's infrastructure department and they are going to be deploying it. I forget what exactly it was that they using it for but it's being put on a IBM server grid if my memory serves correct. ... I ended up going off on a tangent about Portage and USE flag optimizations until a co-worker of his reminded us we were not at work :lol:

Next to improvement in Portage I think the biggest hold up is official documentation lagging behind the latest packages and methods used to configure applications though a frozen tree would help in that regard.
In the realm of commercial acceptance; with more and more usage by academics organizations that boarder or cross into the concept of a standard enterprise cooperation the pure enterprise cooperations will begin to pay attention. It's one of the reasons computers have proliferated to the extent they have.

EDIT: obviously I voted yes :wink:

_________________
Gentoo! Emerge from the package.
Custom Desktop: Dual Dual Core AMD Opteron 280 @ 2.4Ghz - 4Gb ECC RAM - Nvidia GeForce 7950 GT 256Mb
Dell M1210 Laptop: Core2 CPU T7200 @ 2Ghz - 2Gb RAM - Nvidia 7400 Go - 3945ABG Wireless

sanman · n00b Joined: 17 Mar 2007 Posts: 17 Location: UK

eccerr0r · Posted: Thu Aug 16, 2007 12:23 am Post subject:

I voted 'no' because it looks like the Gentoo community is still in flux, in-fighting of where it needs to go. There are a lot of the community that wants "latest and greatest at any cost" which doesn't work if it breaks backward compatibility.

Until the in-fighting stops, and the realization that some way to ensure stability (i.e. NO MORE expat issues, or a way to _clearly_ make sure that stuff like expat never happens again), they'll never choose support for Gentoo.

The versioning system needs to come back to help support (i.e., longer product cycles. Gentoo is virtually a new product every time a new ebuild comes out.)

My opinion on what makes a commercially viable solution for companies to support:

* widespread popularity (ease of installation and eye candy/user interface is part of this)
* Stable versions (not just stable versions, but backporting of bug fixes in later versions without feature backporting unless it's guaranteed not to break anything)
* limited number of possible configurations that could lead to breakage. Yes, that includes a forced version that includes CFLAGS being hard coded and must not ever change.

Sometimes two out of three are enough, but there's a reason why RedHat, SuSE, and even Debian/Ubuntu has won over Gentoo in the past as well as currently.

The ideal case is that they supply source of course, and we'll just tailor it to how we want it

Maybe Gentoo just doesn't want to go that route, and just stay away from things that want stability...
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

tutaepaki · Posted: Thu Aug 16, 2007 8:25 am Post subject:

I think there is another important requirement before any large enterprise will deploy any OS...

A large, and established vendor to support the OS!
Running any important software without a vendors to support, (and blame!) is one of those career limiting moves which no-body wants to undertake.

steveL · Posted: Fri Aug 17, 2007 8:48 am Post subject:

Well I found this thread linked from this blog [yes i know referrer spam] which is linked from drobbins' blog post on why we should all jump ship to Sabayon (yeah right.)

The first one is my first post to dev m-l which made me chuckle, and it's about a frozen tree with every release (I was asking about a minimal b0rkage tree and this was already under development internally for the same purpose, but time constraints meant it wasn't happening) which would be a basic requirement for the stability an Enterprise Edition would need. I was thinking: even if the devs don't have enough time, there's no reason we couldn't start working on it based on a portage snapshot. If Gentoo are going to do it, it would need volunteer users to help out in any case, and I am sure if enough of us got together, relEng would gladly let us use the scripts mentioned (we can get em up to speed if needed) to implement the proposal. (If they don't have a team on it yet, ofc. I ain't proposing it if no-one else wants to do it.)

What does anyone else think? Presumably you care about enterprise support if you're reading this: can you put in say a couple of hours a week?

96140 · Retired Dev Joined: 23 Jan 2005 Posts: 1324

eccerr0r · Posted: Fri Aug 17, 2007 2:45 pm Post subject:

This is also another sticking point: current Gentoo devs would rather upgrade than backport. If package x version 1 has a security bug and version 2 as it fixed but breaks package y, we'll upgrade both package x and package y.

However package y also breaks something else due to its file format changes. Package z is now needed to fix problems with package y. So in comes package z.

Well, this doesn't exactly capture what I wanted to capture, but it's an internal fight of trying to be bleeding edge versus staying back and not getting paid to staying back a few versions and backporting security fixes into x version 1, so we don't have to fix package y in the first place.

I do have to say, things like kernels, the "enterprise approved" behavior does happen but the cycle time is still way too short. And of course as time goes on, the amount of effort it takes to backport gets harder and harder for that one particular person doing the backport.

The people who do have Gentoo "working" in the enterprise are either spending a lot of time chasing package y and dealing with z, or simply let package x go obsolete. Or perhaps manually fixing it. But multiply that out by the number of machines and possibility of conflicts, it grows into a hairy mess fast as it simply takes time that could be better used for other stuff.

Gentoo is now "Have it OUR/MY way or the HIGH way." Enterprise has money to throw at it for stability, and hence prefer the RedHats and SuSEs that tailor to THEIR way.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

turtles · Veteran Joined: 31 Dec 2004 Posts: 1655

I voted Yes because we use Gentoo on our servers to run a enterprise. I dont now of any enterprise applications that cannot be run in Gentoo given a enterprise level budget.

Pro's
1) Gentoo support for Postgresql is top notch.
2) Apache works great.
3) Egroupware and others are also supported by Gentoo.
4)Forums
5)Moving tree
Cons:
1) http://www.gentoo.org/proj/en/glep/glep-0042.html implementation is the main thing holding Gentoo back. Once we know what will happen BEFORE we emerge, we don't mind moving with the tree.
Expat 2.0.1 without pre-warning or a way to prevent a unplanned buggy revdep-rebuild process Gentoo will remain viewed as less than optimal soloution at best.
2) Implementaion of --as-needed flags would improve ones ability to move with the tree. Seems like they are still buggy and not a priority.
3) Lack of documentation on making custom sets. I would like an emerge