CentOS and commercial linux

#1

Haven’t seen any threads on CentOS yet.

CentOS is a vastly popular, free rebuild of the (debatably-expensive) RHEL, a commercial Linux distribution with a long release cycles. The primary benefit of these is you can run them for long periods (10+ years) and still receive security patches without having to upgrade and potentially break your server or appliance.

Due to GPL, people can mostly rebuild RHEL and distribute it for free, although it requires getting access to the source, figuring out how to compile it as well as removing trademarks. This puts RHEL in a difficult spot as a commercial concern, and there is some controversy around tactics they use to discourage people from doing this, although ultimately since they are big contributors to Linux everyone kinda overlooks it.

It turns out it requires actual work to do a free rebuild, and volunteer project CentOS struggled to put out releases in a timely fashion. In the middle of this, RH made a surprise move to buy the CentOS project. There was some concern at the time about a conflict of interest, but it was also thought to be in RH’s interest to have a free offering that would help people grow into RHEL commercial customers. For awhile, that’s pretty much what happened and things improved for both projects.

More recently, RH killed the CentOS project. In short, a lot of linux sysadmins unexpectedly have to migrate from CentOS in the next few months (well before the 10+ year cycle they expected). To put it mildly, this has caused a lot of controversy.

Particular events of note:

Some questions for this group might be:

  1. How viable is the commercial linux business model, given the inevitability of rebuilds?
  2. alternatively, how viable are free rebuilds given that they’ve struggled to keep up with RH as independent projects?
  3. What are the implications of vendors like Oracle selling a rebuild that undercuts RH on price?
  4. Are 10+ years updates as good as it used to be? I’ve seen this come up a lot in the context of containers or immutable infrastructure filling this role.
  5. To what extent is this part of the recent trend of business model issues in oss projects, mongodb, cockroach, mapbox etc? Obviously you can’t relicense the linux kernel like you can in small projects, but some things seem similar.
  6. What do we think about the RH model of selling binaries and disconnecting customers who redistribute source?
1 Like
#2

I get the feeling that while Red Hat continues to do a fair bit of business on early-2000s-style RHEL deals, they see their focus and growth potential elsewhere. In other words, this is legacy business. And ever more an input to the services and other “solutions” they offer with more gusto.

#3

I don’t think this is a business model trend — more like the commodification of the operating system layer.

As far as server operating systems go, I’m mostly not going to want to have to touch the OS or maintain it at all — with various PaaS or Serverless other options so I don’t have to care.

For personal use of server or desktop, free & non commercial offerings will be fine. Which loops around to, is there enough support for maintainers of those offerings?

I backed a recent ElementaryOS funding drive — got the sticker and the mug.

2 Likes
#4

Good links and questions.

How viable is the commercial linux business model, given the inevitability of rebuilds?

Viable for the those vendors for the foreseeable future.

alternatively, how viable are free rebuilds given that they’ve struggled to keep up with RH as independent projects?

The criteria for success (and → viability) is not the same. (F/L)OSS has shown chronic capabilities in terms of living outside the uterus of proprietary tech, despite the fierce, self-preservation instincts of capital.

What are the implications of vendors like Oracle selling a rebuild that undercuts RH on price?

Standard capitalist bun fight. We are all bystanders.

Are 10+ years updates as good as it used to be? I’ve seen this come up a lot in the context of containers or immutable infrastructure filling this role.

got links?

To what extent is this part of the recent trend of business model issues in oss projects, mongodb, cockroach, mapbox etc? Obviously you can’t relicense the linux kernel like you can in small projects, but some things seem similar.

Oss carries the seed of this trait, obviously.

What do we think about the RH model of selling binaries and disconnecting customers who redistribute source?

It’s a flex that fails, IMO. A court case of this nature to settle it may become more likely with time?

#5

I am really interested to know why should anyone go with this when Debian or Ubu... | Hacker News Is a random thread I saw.

For myself, 10 years is a lot, I prefer 3-5. But maybe 10 years ago I’d have wanted 10.

It’s not a legacy business, just currently out of fashion. The trouble with a lot of the “cloudy” offerings is they’re either proprietary or effectively proprietary. at some point the cost becomes an issue and you try to leave. see The Cost of Cloud, a Trillion Dollar Paradox - Andreessen Horowitz for more of an MBA perspective.

I dunno, RH writes a lot of the patches I run.

#6

Okay, Thanks.

Agree, looking beyond 5 years isn’t a great strategy.

I think that’s being a bystander in the whole linux infrastructure bun fight, unless you contribute I s’pose… but then even core infrastructure developers are just pawns, no?

1 Like
#7

BSDs nowadays also has good support and their licensing is more free than GPL.

I find it very easy to modify , maintain NetBSD than compiling CentOS

1 Like