VFX Linux Distribution Recommendation Report
VFX Linux Distribution Recommendation Report
TASK FORCE
20
22
TABLE OF
CONTENTS
03 22
Executive Summary Recommendations
06 24
Introduction Further Guidance
08 26
Why Linux? Longer Term Vision
10 28
The Process Next Steps
11 29
Linux Distributions References
21 30
Evaluation Criteria Acknowledgements
Studios urgently need to choose where to turn next, and there is a real risk of
fragmentation if many different choices are made which would be harmful to both
studios and software vendors. There is consensus that coordinated urgent action
across the community is needed to enable Linux to continue being the operating
system of choice for most studios.
Earlier this year, the VFX Linux Task Force was launched with the single goal of
recommending a Linux distribution for VFX and animation studios to align on for use
on workstations. This report is the output of that task force after several months of
research, outreach, and discussion with Linux distribution vendors, software vendors,
and other members of the community. Although the report is focused on workstations,
the recommendations presented here are likely to inform choices made regarding
render nodes and backend servers too.
While many studios have good reasons for using Linux, Windows and macOS are both
seeing wider adoption as an OS for general VFX use and have specific use cases
prompting most studios to use a mix of operating systems. This report is aimed at all
studios, as the whole community benefits from having Linux as a healthy sustainable
choice even if they choose not to use much, or any, of it.
The urgency and scale of work needed across the community imposed some
constraints on the choice of distributions to consider, which is elaborated on later in
this report. While all options were considered, ultimately, the distributions that were
more specifically assessed were AlmaLinux, Red Hat Enterprise Linux, Rocky Linux, and
Ubuntu.
For artist workstations running Linux, it is strongly recommended that all VFX
and Animation studios deploy Red Hat Enterprise Linux (RHEL) 9.x or one of
its binary compatible rebuild distributions such as AlmaLinux 9.x or Rocky
Linux 9.x in 2023, and no later than June 2024 for those still running CentOS
Linux 7.
While there were many factors that led to this recommendation, one of the strongest
was that there is already a great deal of infrastructure, tooling, and expertise
around RHEL / CentOS Linux in the community. While there are benefits to other
distributions, they were collectively not compelling enough to build the critical mass
needed for so many studios and vendors to agree to retrain and retool in such a
short time.
In the longer term, the community should pool know-how and effort to elevate the
professional artist workstation experience on Linux to be at least on a par with, and
hopefully exceed, Windows and macOS. New virtualization and containerization
technologies should enable more flexibility, and increased interoperability, so that
studio workstations with different operating systems can more easily co-exist with
each other. Software vendors and studios should work together to ease the
adoption of Linux for studios that want to increase its use on workstations. Finally,
better community coordination can help increase the ease and frequency of
software updates. This would help the whole community to adopt new capabilities
more quickly, and use more recent releases that offer better security, performance,
and functionality.
Ultimately, these recommendations and the next steps outlined in this report are an
important step on the journey of securing the future of Linux as a sustainable, healthy
platform for studio workstations, providing a more easily adoptable solution that will
eventually deliver the best artist experience on professional graphics workstations.
The early years of the industry were dominated by Silicon Graphics before Linux
became the natural successor to SGI’s IRIX. Two decades later, Linux is still the primary
OS used on artist workstations at most studios, but there is uncertainty about the future
after the decision by Red Hat and the CentOS Board to replace CentOS Linux with
CentOS Stream, a continuously updated distribution that may not fit the requirements
of our industry. At the same time, macOS and Windows are more powerful and
capable than ever.
In October 2021, the Visual Effects Society’s (VES) Technology Committee conducted
the first Studio Platform Survey, the results of which were published in a report in
January 2022. Among other insights, the report confirmed that:
As a result of these findings, it was clear that coordinated urgent action across the
community is needed to enable Linux to continue as the workstation operating system
of choice for most studios. Given its impending end of life, including the end of critical
security updates, it is imperative that studios move quickly away from CentOS Linux.
The community also must avoid the fragmentation from each studio choosing to move
to different distributions, as that would increase costs and be difficult for software
vendors supporting their products on Linux. Distribution fragmentation would also
create additional burdens for industry-specific open source projects, as well as for in-
house teams consuming these projects or developing their own projects with an eye
towards collaboration with other studios.
This report is the output of that task force. It explains the process, research,
assessment, and finally the recommendation for which distribution to use. It also lays
out a longer-term vision and proposed next steps toward the goals of:
This report and its recommendations are very much focused just on Linux distribution
selection for artist workstations at multi-user VFX and animation studios as this is where
most urgent action is required. These recommendations are not aimed at individual
artists, who may value being on the latest software over stability or interoperability, or
at other uses of Linux.
It is valuable to understand why Linux is preferred for this use by most studios, and why it is
worth coordinated effort across the community to sustain it. The primary reasons tend to be:
Flexibility - Open source, standards support, and a large community means Linux is
flexible and there is a huge library of code available, plus the opportunity to customize
the OS to meet specific needs.
Vendor independence - Linux supports a wide array of products which provides the
ability to tailor systems towards specific problems, and not be locked into product life
cycles with software-enforced obsolescence.
Performance - Full access ACTIVITY
KEY INDICATOR to the kernel allows for tuning
/ PROJECT to /optimize
DATA OUTCOME for high-
performance computing applications like simulation and rendering. With large-scale
deployments of render farms and expensive workstation hardware, even small
performance gains can provide high value.
Network transparency - Linux allows users to access resources without the user needing
to know, or even be aware of, whether the resource is local or elsewhere on the network.
Most studios make heavy use of this feature for storage using NFS.
Supportability - The fully open nature of Linux means it is possible to investigate and
potentially solve any problem without the necessity of relying on a vendor that may not
understand the problem. A studio with access to technical expertise can potentially fix
blocking issues affecting its production schedule without having to wait for a solution from
an external vendor.
Licensing - The option of no-cost installs, avoiding license management complexity and
constraints to scaling up infrastructure. This is especially important when deploying to
public clouds, where operating system licensing costs can fundamentally change the
economics of what is feasible and what is not. NB: Some Linux distributions do require a
paid subscription for access to support and services.
Again, Windows and macOS are both great options too and most studios use them in
addition to Linux, with a good proportion of studios being primarily Windows-based.
This report is aimed at all studios, as the whole community benefits from having Linux as
a healthy sustainable choice even if they choose not to use it.
CentOS Linux 7 is approaching its end of life on June 30th, 2024 but studios will
need to have been working on a replacement long before that in order to
complete the transition in time, hence the need for some urgency.
The Task Force consisted of 10 people, all of which are lead systems engineers or
technology leaders from a mix of small, mid, and large VFX and animation studios,
with representation from studios that are predominantly Windows as well as
primarily Linux-based.
The group met weekly throughout the process, often inviting industry partners to
hear their perspectives and discuss the various options. Among others, the group
met with:
The publication of this report signals the end of the VFX Linux Task Force. The next
steps outlined below will hopefully germinate new groups to build on this report’s
recommendations.
Furthermore, some studios and software vendors require an option that gives them
the ability to pay for support, and/or provides a means to financially support open
source work in order to help with sustainability, and perhaps have some additional
influence in competing over priorities with other industries.
Neither Debian nor Mint have options to pay for additional support or services, so
they did not qualify for consideration. Kubuntu was rolled into the overall assessment
of Ubuntu.
Although AlmaLinux did not appear in the Studio Platform Survey Report, as another
RHEL-based rebuild with a strong community focus, it objectively meets all of the
selection criteria along with Rocky Linux. Since both AlmaLinux and Rocky Linux are
still relatively new, having more than one option here helps minimize the risk of
either choice so the decision was made by the Task Force to include it in the
assessment.
So, the final distributions that were assessed were AlmaLinux, Red Hat Enterprise
Linux, Rocky Linux, and Ubuntu.
Adequate experience and expertise with the distribution in the VFX and
animation studio community
Availability of optional paid-for support and/or services
Enough stability, and a long enough support lifecycle, to meet the needs of
studios and their software partners that often work on multi-year projects
Conversely, the distribution needs to move fast enough, and consistently
enough, to enable studios to adopt new technologies without unpredictable
delays
Licensing needs to support studios using a mix of open source, 3rd party
commercial, and proprietary in-house software
Prompt, robust response to security vulnerabilities
First-tier driver and library support from commercial GPU vendors.
The reasons for this are not only that it is so much work to set up and sustain such a
distribution, but also that software vendors are much more likely to support their
products on a widely used Linux distribution that reaches a broad market than a
specialist "VFX Linux" distribution that targets a smaller proportion of their customer
base.
Whenever this topic has come up in the community, the conclusion has always been
that leveraging an existing, widely used distribution is much more likely to be
successful, and that is still the case today. Instead of a custom distribution, it is
expected that community efforts may well result in a bundle of additional packages
and configuration settings that layer on top of an existing distribution to provide a
more consistent artist experience that is better tuned and easier to set up.
Although it is not available free of charge, RHEL is still a true open source
distribution. While the myriad available of open source licenses differ in their legal
requirements for publishing source code, Red Hat just uploads the source code on
release for every software package that makes up RHEL to https://git.centos.org/
where it is publicly available.
While the source code is free for others to use, Red Hat's branding and trademarks
are not. So what the CentOS project did (and what other downstream distributions
like AlmaLinux and Rocky Linux do today) is take the source code that Red Hat
publishes and re-compile it with a minimal set of changes to remove anything
trademarked by Red Hat (and replace it with their own branding). They then make
the result available free of charge.
This process does not only apply to the initial release. Every time Red Hat updates a
software package (either for security, bug fix, or enhancement) the cycle repeats.
This is why both AlmaLinux and Rocky Linux can claim to be "1:1" or "bug-for-bug"
compatible with RHEL.
The "bug-for-bug" statement is important to take Please note that not all
note of: if a bug or security vulnerability exists in downstream distributions are
created equal, and some are
RHEL, it exists in any of the rebuild distributions
intentionally not 100%
until it is fixed by Red Hat and the rebuild cycle
compatible. For example,
repeats. Oracle Linux is another well-
established RHEL rebuild
Thankfully, modern CI/CD and automation have although it did not qualify
greatly improved this process and both for inclusion in this
AlmaLinux and Rocky Linux are setting new assessment. Although Oracle
records for reducing the gap, with many updates claims 100% application
binary compatibility with
generally being available within a day of
RHEL, Oracle Linux is not
release by Red Hat. This release gap can never
quite a 1:1 rebuild. Because
quite be zero and that may be a major there are changes in some
consideration for some when choosing between core packages such as glibc,
RHEL or a downstream rebuild. it may not be “bug-for-bug”
compatible with RHEL, or
offer identical performance
characteristics.
Red Hat was acquired by IBM in 2019, and early concerns have been allayed by
RHEL’s continued development and success, and Red Hat’s strategic importance to
IBM. This makes RHEL a relatively safe and secure bet in terms of its longevity and
sustainability. In terms of the sustainability of the RHEL Workstation product
specifically, it is used widely across multiple industries including automotive,
aerospace, semiconductor design, geological research, and scientific visualization
among others so its future seems secure.
With the launch of RHEL 8, Red Hat defined new life cycle rules to make consumption
of the platform simpler and more predictable. They are as follows:
New minor releases every 6 months for the first 5 years with X.10 being the final
minor release.
Major releases of RHEL will still be supported for 10 years. There are two phases of
support, Full Support, and Maintenance Support. During Full Support (first five years)
there will be new feature enhancements, hardware enablement, etc, whereas
Maintenance Support (the remaining five years) will be for just Red Hat rated
Critical and Important security and bug fixes.
From: https://en.wikipedia.org/wiki/Red_Hat_Enterprise_Linux
For more information on the RHEL life cycle, please review the RHEL Life Cycle
knowledge base article.
A majority of studios choose to use a distribution that is available at no cost, but some
do pay for RHEL subscriptions for their workstations, servers, and render farm. Although
annual investment is required for these subscriptions, in return the studios benefit from
expert Linux support and the ability to influence the RHEL roadmap to better meet the
studio's needs.
AlmaLinux
AlmaLinux is a 100% binary and “bug-for-bug” compatible rebuild of RHEL, so it has
the same technical capabilities and product lifecycle.
Rocky Linux
Rocky Linux is another 100% binary and “bug-for-bug” compatible rebuild of RHEL,
with the same technical capabilities and product lifecycle.
Its founders include Greg Kurtzer, one of the original co-founders of the CentOS
Linux distribution, and its goal is to learn lessons from the CentOS Linux experience
to create a more sustainable production-ready downstream version of RHEL.
Rocky Linux is owned by The Rocky Enterprise Software Foundation (RESF), a Public
Benefit Corporation (PBC) formed in Delaware. The RESF is backed by an advisory
board of trusted individuals and team leads from the Rocky Linux community. Ctrl IQ
Inc (CIQ) is the founding support and services partner of Rocky Linux.
Rocky Linux is also early in its life but has already been making a big impact. It has
built an impressive list of sponsors, and CIQ recently received a $26M round of
funding, so this seems to be good progress towards long-term financial security.
Releases of Rocky Linux have reliably followed RHEL releases and updates, although
the recent Rocky Linux 9 release came 8 weeks after the RHEL 9 release. This was
due to Rocky Linux 9 being the first release to be built with a new community-
developed and open-source cloud-native build system. As per their press release, “a
primary goal in developing the new build system was assuring that new versions of
Rocky can be released within one week after each Red Hat Enterprise Linux new
version release”, which bodes well for future Rocky Linux releases and updates.
Canonical offers optional commercial support services for Ubuntu and all editions of
the distribution are freely available, with support for 5 years for the LTS releases.
Extended 10-year support is available for LTS releases if purchased.
Similar to Red Hat, the latter 5 years of extended support are limited to security
fixes.
Another particular strength of Ubuntu over RHEL is the strong support for various
desktop environments via the Ubuntu flavors. While they are owned and developed
by the community and not Canonical, they benefit by being fully backed by the
Ubuntu archive for packages and updates making them well supported. The official
flavors that are based on desktop environments currently include Kubuntu (KDE),
Lubuntu (LXQt), Ubuntu Budgie, Ubuntu MATE, and Xubuntu (XFCE).
Technical
Velocity/Reliability - Each distribution strikes a different balance between
reliability & stability, and leaning into the latest releases offering new features
and capabilities. While individual artists and developers may prefer high-velocity
progress over stability, studios and software vendors tend to lean more towards
stability to ensure enough interoperability and reliability to support their
businesses.
Adoptability - The ability for studios and vendors to adopt the distribution quickly
given the urgent need to replace CentOS Linux.
Lifecycle - How long each release is supported for, how often releases are made,
and how quickly downstream releases follow the upstream releases.
Future features - Confidence in the distribution provider addressing VFX-specific
needs such as Wayland support for complex DCC tools, and support for
specialized hardware such as professional HDR, wide gamut and XR displays.
Commercial
Engagement and influence - The level of engagement the distribution provider
has with VFX and animation community, and their level of partnership and
influence with key technology partners such as HP/Teradici, Autodesk, Epic
Games, Foundry, Nvidia, SideFX, Quantum, Wacom etc.
Support service - The availability of optional additional support services,
potentially for an additional cost, for studios that need or want it.
Governance model - Whether the governance model of the distribution or
distribution provider is healthy, and aligns with the interests of the community.
Financials - Financial security of the distribution provider.
Engineering resources - The provider's level of engineering resources to sustain
an appropriate level of service and innovation.
Sustainability - Confidence in the long-term sustainability of the provider or
distribution.
Industry knowledge and experience - Existing know-how and experience with the
distribution within the VFX and animation community had to be a consideration.
While freedom of choice is one of the strengths of the Linux ecosystem, in the
case of the VFX and animation industry it is more important at this time that the
community aligns on a common distribution to lower the barrier to Linux adoption
by more software vendors and VFX studios. It also allows the community to have
more focus on a path forward as we manage the transition from X to Wayland for
a community with such graphically complex tools.
The main reasons why RHEL and its downstream distributions were selected are:
The reason for specifically recommending RHEL 9 for studios is that RHEL 8 is
already approaching the end of its 5 years of Full Support in 2024, and will then
enter its 5 years of Maintenance Support. Studios adopting RHEL 9 will give them
the ability to benefit from new functionality as it is released, as well as more
leeway before the next major upgrade is needed.
Giving software vendors the option of building on RHEL 8 allows them to be more
inclusive of supporting older distribution releases still used by some of their
customers. Software can be built on RHEL 8 in a way that enables seamless
support on RHEL 9, as well as including support for some other distributions that
do not include a version of glibc as recent as the one included in RHEL 9. A
relatively simple way of building on RHEL 8 while gaining access to newer
compilers and toolchains is by using the Red Hat Developer Toolset product.
Despite making a recommendation for RHEL and its rebuilds, the Task Force would
like to take this opportunity to acknowledge that Ubuntu is a very strong
contender and the Canonical team has been an excellent and supportive partner
through the research and assessment process. We encourage the community to
continue to work with Canonical with an eye to making Ubuntu an even stronger
contender for professional graphics workstations in the future. In the meantime,
although it is not our recommendation, we do expect some studios may still
choose to use Ubuntu on some workstations for specific use-cases. In the longer
term, it would be very beneficial to see equal support from all DCC vendors for
both RHEL and Ubuntu and so this is a secondary recommendation once the
primary recommendations above have been addressed.
In addition to the recommendations laid out above, the Further Guidance and
Next Steps sections below should be considered a secondary set of
recommendations from the VFX Linux Task Force.
One example of why some studios elect to pay for RHEL is that it secures agreement
terms that provide accountability for the governance of the supply chain. A Linux
distribution is the end result of a complex set of inputs that are difficult to manage
securely, and vulnerable to supply chain attacks. A RHEL subscription provides legal
commitments, patent protection, and accountability from the distribution provider
that are not currently readily available from more community-driven downstream
distributions.
Overall, it would be to the community’s benefit for more studios to financially support
the Linux ecosystem since it would help ensure that development efforts go towards
the features that will be most impactful for VFX and animation, and help to create a
healthier and more sustainable ecosystem.
An individual minor release only receives about six months of updates, and this support
ends once the next minor release is available. "Pinning" support to a specific minor
release therefore puts studios that stick to that release at risk. While extended minor
release support is available (for an additional cost and only with RHEL), the vast
majority of DCC users do not have access to it. Given the security compliance
requirements that the studios have to commit to, supporting future minor releases is
essential to enable studios to apply critical security patches without falling out of
support. Studios following the required security practices will be installing updates on
an ongoing basis and by the time a new minor release comes out, they will already
have installed a significant number of the updated package versions rolled up in the
new minor release.
When specifying system requirements, vendors should use minimum versions and not
static versions. For example, "RHEL 9.2+", "RHEL 9.2 and later", or simply "RHEL 9".
Also, vendors should specify something along the lines of “RHEL includes Red Hat
Enterprise Linux and compatible downstream or rebuild distributions including Rocky
Linux or AlmaLinux".
Specifying the supported Linux distribution releases more broadly in this way should
not increase the support burden for vendors, but it will make a much more
straightforward on-ramp for the many studios that need it and it is expected that this
would translate into increased adoption.
While this is just a recommendation, it is expected that many studio customers will
make this a requirement of their software vendors.
In the longer term, the community should pool know-how and effort to elevate the
professional artist workstation experience on Linux to be at least on a par with, and
hopefully exceed, Windows and macOS. New virtualization technologies should
enable more flexibility, and increased interoperability so that studio workstations
with different operating systems can more easily co-exist with each other. Software
vendors and studios should work together to ease the adoption of Linux for studios
that want to use more Linux on workstations. Finally, increased community
coordination can help increase the ease and frequency of software updates which
helps the whole community to adopt new capabilities more quickly, and use more
recent releases that offer better security, performance, and functionality.
Aligning on compatible Linux distributions will make it easier for the community to
pool their efforts and benefit from building on a common foundation. For too long,
the knowledge of how to build powerful Linux-based studios has been locked up in
the silos of individual studios. It is time for the community to come together to pool
that knowledge and make it available to all, creating a shared repository of
configuration guides, best practices, and tutorials, that are continually updated and
improved for the good of all.
Such a knowledge base would enable more studios to adopt Linux, which in turn
would encourage more software vendors to support Linux as a first-class platform for
their products. A more standardized, and better understood, Linux platform for studio
workstations would also encourage hardware providers such as Intel, AMD, and
Nvidia to continue to provide first-rate open-source contributions, such as drivers
Future use of modern containerization approaches for artist tools will simplify the
support burden, and enable more flexibility when it comes to customization of the
underlying platform to meet different studio needs. It is likely that eventually these
components will also be integrated with package management tools to make it
easier for people to install, manage, and run them. All this progress will require
research, expertise, and resources. Progress will be much more rapid if this is done in
the open as a collaborative effort between studios and their technology partners.
With the increasing pace of technical progress, and the need to keep up to date
with security patches, it is anticipated that studios will need to invest in upgrading
their systems on a more regular basis, with major OS upgrades happening on a 3-
year cycle. More frequent upgrades could make application compatibility more of a
challenge, which is another reason why collaboration across the community is
essential. By similarly promoting an overlapping pattern of support within the VFX
Reference Platform for software updates, the critical libraries for the community can
provide either ABI compatibility for that maintenance envelope to enable backward
compatibility, or software configuration options and support for having multiple
versions of the software installed side-by-side. Combined with that, the community
should also consider whether there should be a maintained repository of packaging
recipes for the package deployment systems (rpm, deb, etc.) as well as environment
management tools such as Rez, and perhaps a repository of pre-built packages.
In summary, we see a path forward for studios to work together to create a robust,
sustainable workstation platform that offers the best choice for a premier artist
experience, and the preferred deployment platform for high-end digital content
creation tool providers. This can be achieved by sharing knowledge and providing
some consistent software deployment recipes and practices. This will assist in
building and deploying systems that leverage the power of open source software to
deliver the open standards and foundation which enables everyone to deliver the
next generation of creativity and innovation.
The authors would like to thank contributors from the following organizations for
their partnership, support, and feedback throughout this process:
AlmaLinux Foundation
Academy Software Foundation
Autodesk
Canonical
CloudLinux
CIQ
Epic Games
Foundry
Red Hat
Rocky Enterprise Software Foundation
Side FX
VES Tech Committee
VFX Linux Task Force
VFX Reference Platform Working Group