Now that we are approaching the 1.2.0 LTS release, it's time to make a big announcement. Perhaps we should have made it earlier, but we've been too busy coding.
There are two distinct categories of VyOS users. The first category is people who want the latest features even at cost of stability. These are mostly networking geeks who run it in their home networks and network labs and open source developers, though some businesses are also happy with this approach. The second category is people who need or want stability. There are of course people who want both too, but we have to accept that these goals are contradictory at least some of the time.
This is common for all software projects, but with VyOS, more people seem to belong to the second category. Every once in a while someone asks on the channels and the forum whether they can update an extremely outdated VyOS (or even Vyatta Core) version to the latest release. We also hear from people frequently that they are not going to even try 1.2.0 until it reaches the stable status.
It's quite obvious by now that the single release model is not fitting for this situation. With 1.2.0, people who contributed new features had to wait a long time for their code to appear in any non-nightly build image because other people, mainly the maintainers, have been working on tearing the codebase off of the outdated Debian release, reworking the foundations, and hunting bugs. This is frustrating for contributors and even created the appearance of the dead project at some point. It also means that new code doesn't get to people enthusiastic to test it nearly as fast as it should.
On the other hand, while we have an active community of people who send us patches and report bugs, the community is very small compared to the entire user base. The number of people who contributed patches is easy to measure so I did it out of curiosity: the largest submodule, vyatta-cfg-system, has less then 50 unique names in its commit log starting from the project start date, which means less than 50 people contributed to it in five years. The number of active testers isn't much higher. For comparison, the 1.1.8 images get thousands of downloads every month and the wiki documentation gets thousands of views too. To make it easier for people to contribute, there's a lot of work to be done: reworking the foundational libraries, getting rid of poorly written legacy code, and so on, and focused effort (which means human-hours) is required to break the cycle.
Maintaining stable releases is one of the hardest parts, but that burden falls on a disproportionately small number of people, while most of the business users who are the main consumers of stable releases do nothing to advance the project. This is not a healthy or sustainable situation. Someone needs to pay the bills.
The new release model
First, there now will be two release lines: rolling release and long-term support releases.
The rolling release images will be (roughly) monthly snapshots of the "current" branch, with all latest pull requests merged in. They will be tested to successfully boot and load a sample config. The target audience is the people who want the latest features (even if they are not working perfectly yet). People who send us pull requests can be sure their contribution will be available to themselves and willing testers in a reasonable time. Since in VyOS it's easy to revert to the previous version if something goes wrong, the rolling release should be good enough for non-critical production use, since you can always go back to a working version at the end of the maintenance window and report the findings.
The long-term support versions will be maintained for at least two years from the release date. They will undergo extensive testing by the maintainers, and receive backported bugfixes and security updates until they reach an EOL, with a possibility of receiving extended support by special agreement. It is meant for enterprises and service provider users.
Unlike the rolling release images, binary LTS images will be only available to people who help the project move forward, either by contributing their time or their money. We are not going to compromise the free software ideals: you will always be able to build LTS releases yourself.
The LTS images will be available at no cost to all people who contribute to VyOS. Every kind of contribution counts:
Writing code
Testing release candidates and rolling/nightly build and reporting bugs
Writing documentation
Promoting VyOS in blogs, social networks, and at conferences
Everyone who contributed before the release of 1.2.0 LTS version will get a perpetual subscription. People who will have joined later will need to be active within the last year to maintain their subscription (the required activity level is yet to be determined, but it will require substantial and non-trivial changes, i.e. not just typo fixes). Companies who allow their employees to work on VyOS during working hours or specifically pay their employees or contractors to work on code that is meant for the mainline VyOS or produces open source integration tools will be able to get a corporate subscription if those employees/contractors confirm it.
We are also happy to provide subscriptions to contributors of all projects that VyOS uses, such as FRR, netfilter, OpenVPN, StrongSWAN, and many others.
Companies who simply want to use stable, long-term support releases without making technical or social contributions to the project will have to purchase a binary image subscription (you pay for access to ready-made images, not software licensing as such).
All money received from the paid subscriptions will be used to fund VyOS development, including paying the salaries of the VyOS maintainers who work at Sentrium, hiring/contracting developers from the community, expanding the project infrastructure, and supporting our upstream projects.
If you have contributed to VyOS, you can register right now using this form. We will post the details of the commercial offer and pricing later, stay tuned.
P.S.
Back in 2013, I said that there will never be a "VyOS Subscription Edition". Technically I lied, but it was said from a very different perspective. At the time we hoped that now that VyOS is open for everyone's contribution, a large contributor community will form and many of the corporate users who used to use the old project will contribute to it willingly, sponsor its development, or purchase support subscriptions; in practice very few of them did. That approach didn't work, but switching our AWS Marketplace offer from free of charge to paid has become the first reliable funding source for the project and already allowed us to add support for more cloud platforms, as well as rework some of the fundamentals of VyOS to make it easier to contribute to.
I hope continuing this line and introducing the rolling release will create a model that is more beneficial for the project and more fair towards its contributors.
Just to clarify, VyOS is neither going to hide the toolchain required to build LTS releases yourself nor going open core. Those part of the original plan do and will stay the same.