|
|
Subscribe / Log in / New account

Introducing Incus

Introducing Incus

Posted Aug 8, 2023 16:34 UTC (Tue) by geofft (subscriber, #59789)
In reply to: Introducing Incus by ringerc
Parent article: Introducing Incus

Characterizing Oracle as a parasite is inaccurate, see also https://lwn.net/Articles/935627/ . They do a lot of upstream kernel maintenance (searching for "oracle" in MAINTAINERS will get you some of it, but also note there are Oracle folks with kernel.org addresses, e.g., Darrick Wong, who is mentioned in the article) and they regularly show up in the list of top contributors, there is Java of course (which Red Hat makes a lot of money off of), they're active in Xorg upstream, etc. I have no sympathy for Oracle as a company (and this is somewhat personal, they acquired a startup I interned at and promptly dropped support for everything except Oracle Linux), but this characterization is unfair to the sizable number of upstream developers who work at Oracle.

If you want to talk about parasitism and the business motivation for Red Hat's decision, blame the bug-for-bug compatible rebuilds, which allow large companies to run a small percent of their machines with actual Red Hat, the rest of their machines with a binary-equivalent rebuild, and report issues through Red Hat's support contract regardless of which machine it's on. This completely breaks Red Hat's business model, which is that they'll provide you expert engineering and support services in proportion to how much you use their product. If there are tricky locking bugs with a one-in-a-thousand chance of hitting them, and you want ten of them fixed, it makes economic sense for Red Hat to do it if you pay them for ten thousand support contracts; it doesn't make economic sense if you pay them for five and you claim you ran into all of those bugs on five machines.

Oracle Enterprise Linux (which doesn't even default to the Red Hat kernel) does not break the business model: it just competes with Red Hat because they provide their own engineering and support services which they also bill per-machine - and SUSE's Liberty product does exactly the same thing!

Of course, the rebuilds are entirely within their rights to be bug-for-bug compatible rebuilds and the customers who use them this way are mostly within their rights to do so (they're certainly within their rights to deploy the rebuilds; whether they can use Red Hat's support contract to have them fix issues on a large fleet that were reproduced on a small fleet is debatable), because it's open source and the entire idea of open source is that there's no such thing as a parasite.


to post comments

Introducing Incus

Posted Aug 8, 2023 23:06 UTC (Tue) by Wol (subscriber, #4433) [Link]

> because it's open source and the entire idea of open source is that there's no such thing as a parasite.

What do you mean there is no such thing as a parasite? A parasite feeds on its host without contributing back (as opposed to a symbiote, where it's debatable who is feeding on who).

People who charge for support, pocket the money, and then expect someone else to actually provide the support can only be described as parasites. They are seriously damaging to the host.

Freeloaders aren't really a problem, because what's wrong with taking what's freely offered? As above, it's when people are given an inch, and take a mile, that we have a problem.

Cheers,
Wol

Introducing Incus

Posted Aug 11, 2023 7:40 UTC (Fri) by joib (subscriber, #8541) [Link] (1 responses)

I wonder if, longer term, these kinds of issues mean that RH (and other companies with RH-esque business models) will have to figure out some other way to charge for their services than per-OS-instance licensing. Not only because it's easy to circumvent (in the sense of just having one RHEL licensed host that you use to reproduce issues you need help with, as described in the parent comment), but also because the additional hassle of registering hosts etc. is more trouble than just spinning up a rebuild instance with no such requirement.

What such a licensing scheme would be, I have no idea. Per-user licensing suffers from the same issue, in that you can have just one common account that all your thousands of engineers managing your server fleet uses.

Introducing Incus

Posted Aug 11, 2023 7:46 UTC (Fri) by ceplm (subscriber, #41334) [Link]

I would think that working on some better licensing model would be much better reaction to the development of last ten or so years than jerky reactions which seems to be the standard of absolutely everybody involved in this.


Copyright © 2025, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds