Today SUSE announced its intent to do a “hard fork” of Red Hat Enterprise Linux (RHEL), and yesterday Oracle came out with a press release aimed squarely at Red Hat and IBM, and trying to claim the high road in keeping Linux “open and free.” It’s fair to say that the knives are out. It’s not surprising but it is disappointing in a number of ways. I still have a number of other posts in the works but wanted to get some thoughts down on this one before too much time passes.
Oracle and Open Source
If you read Oracle’s press release without any awareness of Oracle’s history with open source, it sure reads like Oracle is a champion of all that’s good and open. It says so many lofty things about Oracle’s contributions to Linux (fair, Oracle has made significant kernel contributions, and Oracle database running on Linux was huge for adoption) and Oracle’s goal for transparency and openness around Linux. That one requires a bit more examination.
Oracle’s foray into producing its own Linux distribution is not one that can be accurately described as having altruistic motives. This will be the topic of another post in the near future but Oracle’s motives for launching Oracle Linux are hardly about a “commitment to Linux freedom.” (Also, a clone of RHEL is not necessary for “Linux freedom.” Oracle could’ve had that by supporting Debian GNU/Linux, and I’m hopeful readers can see that.)
But let’s take a quick survey of some of Oracle’s greatest hits in the open source department:
- Bought Sun and eventually killed OpenSolaris development and ceased releasing Solaris kernel code as open source.
- Had the opportunity to re-license things like ZFS and DTrace under the GPL so they could be ported to Linux or at least code re-used. Chose not to.
- Maintains MySQL as an open core product where enterprise features are held back from the open source releases entirely. RHEL code in some form is available in CentOS Stream. MySQL Enterprise Edition code is simply not available as open source.
- Sued Google claiming APIs are independently copyrightable, a case that would’ve had enormous and unpleasant implications for a whole slew of open source projects and products.
- Maintains VirtualBox as open source, which is great. Offers a bunch of necessary extensions as a proprietary extension pack, which is less so.
Also, it’s very easy for Oracle to champion ideas of openness around Oracle Linux and RHEL: Its Linux offerings are not significant revenue streams for Oracle. RHEL, on the other hand, is a major chunk of Red Hat’s revenue.
Oracle is primarily a proprietary software company that dabbles in open source where it makes sense to support its proprietary offerings. That’s fine, but the company isn’t applying any of these values to its own cash cows. When Oracle, a company that made more than $42 billion in FY22, starts releasing corresponding source code to Oracle Database and its other proprietary software, then it has standing to take potshots at Red Hat and IBM.
Until then, their press release and commentary should be seen for what it is: A substantially proprietary vendor taking an opportunity to kick a competitor when it’s down.
SUSE announces a “hard fork”
Today SUSE announced a “hard fork” of RHEL and has committed “more than $10 million” into this project to “develop and maintain a RHEL-compatible distribution available to all without restrictions.”
Right now it’s a bit light on the details, so it’s hard to determine whether SUSE intends to fork RHEL and keep attempting “bug-for-bug” compatibility with RHEL or if a RHEL release will be a jumping-off point and serve as an industry reference maintained separately. Or any possible variation on the spectrum between those things.
Towards an industry standard Enterprise Linux?
If it’s not clear from previous posts, I’m not a huge fan of the idea that RHEL (and only if it’s bug-for-bug compatible) serves as a de facto standard with Red Hat doing the bulk of the heavy lifting. That’s been the status quo for years, but it’s never seemed equitable or healthy.
Rather than the industry figuring out a way to work together and create a standard reference Linux that is free to all, the default has been to copy RHEL and hope (and complain) that it continues shipping a reference implementation for everybody else. I’ve heard variations on the complaints about Red Hat having too much influence over this or that because it’s doing a lot of the work for years.
Things like the Linux Standard Base and United Linux fell flat because the industry couldn’t work together to sustain and adopt them. Any time Red Hat makes a move to protect its product, it gets complaints about how that’s bad for the community and open source – but the community hasn’t managed to come together and do the work that Red Hat is doing in producing the standard EL implementation.
So we’re at another inflection point. I want to see Red Hat succeed, but I’d also like to see a vendor-neutral distribution that is sustained by the larger community and industry.
Not, as one Mastodon user put it, because it would “hurt IBM.” That’s a terrible incentive for building a community. So is anger at Red Hat, really. Whatever you may think about Red Hat’s business decisions, the company is still doing massive work upstream and has a history of open sourcing all its products to a degree that no other company has matched. Its developers are still working shoulder-to-shoulder upstream with people from Oracle, SUSE, Canonical, and even Microsoft.
Red Hat may have made a huge tactical error here, but it’s their error to make. “The community” has no claim to RHEL as a product, and if the industry has adopted a product as its standard without considering the company behind it could change strategy at any time, that’s a failure on the industry’s part. The only way to ensure continuity here is to step up.
So if that’s what happens, finally, great. If the goal is to create one “enterprise” Linux distribution as a commons that is carried by a much larger group, I’d be all for that. It’s been tried before, and failed spectacularly, but maybe the industry has learned a thing or three in the last 20 years. Maybe.