Contrary to popular belief, there is a lot of work to be done in the CXL ecosystem that can or will not be done by the very recognizable multi-national companies who are investing billions into CXL hardware design and development. Right now, because we are so early in the ecosystem, it does seem to be dominated by those who are making the actual silicon to ‘make it so.’ Anything from processors, retimers, switches, drives, controllers, etc. are all key pieces to making CXL go, but hardware at scale isn’t the only way that CXL will be successful.
I’m a big fan of the 3 things method of approaching “What can I do?”, so here are 3 things you can do if you want to get involved with the CXL ecosystem.
Thing 1: Identify integration and end-user opportunities #
Just because the technology exists doesn’t mean that people immediately understand and know what they are going to use it for. One of the many hurdles to broad use case, mainstream adoption of CXL is the lack of variety in problems it can solve. Because there is currently no relatively painless way to share memory between systems, or have a disaggregated memory deployment there simply hasn’t been much mainstream thinking into “well, what would I do if that was possible?”. Simply the act of thinking about it, talking about it and getting your ideas for what would be a great integration or usecase into the ecosystem is potentially the most valuable way to get involved in the CXL ecosystem this early on.
We talk a lot about the parallels that we feel CXL has with Kubernetes or Cloud Native as a paradigm. What if the end user group for Kubernetes never thought of what was now possible with the orchestration tool, and simply used the tool as is? We definitely wouldn’t have developed as fast and as far as we did. The same problem set lies in front of CXL, it’s true potential can only be realized if the people who develop applications and use the compute resource leverage it in ways that the original hardware manufacturers have no interest in or are simply not thinking about. This isn’t because hardware companies don’t care, it’s because hardware is a hard problem to solve by itself.
- Publish thoughts or wishlists in public on platforms like LinkedIn or blogs or even comments sections of a site like Hacker News.
- Look internally at current architecture deficiencies and look at how CXL can solve them
- Write about problems or dead ends that you run into, so others can think through a solution with new technology
Shameless plug: If you are also just looking to bring in experts who know the specification and it’s capabilities and want to marry that with your current infrastructure or application deployment the experts at Jackrabbit Labs can do deep dive sessions and use-case analysis. Reach out to us if you are interested.
Thing 2: Join the CXL Consortium as an Adopter or Contributor #
This one is pretty straightforward, but a lot of people don’t know that there is a membership tier available free of charge. The Adopter membership allows you to view the spec, attend CXL consortium events and generally participate in the ecosystem without having to do anything other than apply!
Check out more here: https://computeexpresslink.org/join-cxl/
Thing 3: Help code! #
There is tremendous momentum and interest in the CXL hardware world, but the software side has been muted by all of this excitement. If you have or maintain a project that has an intersect with CXL the best thing you can do to get involved is to code! Develop a proof of concept, try some things out, maybe break some things. Without a strong developer community and ecosystem CXL will become merely a tool to optimize BOM costs on large scale deployments, when it’s full potential is to change the way we develop and use applications, a true paradigm shift in computing.
In the coming days we will be Releasing jack, our open source FM API command line tool and alongside that our CXL Switch Emulator. If you are stuck on finding a project to help with, help with Jack or write an integration with Jack to enable your product in the CXL ecosystem. Follow us on github to track the latest releases and changes.
I will also be attending Memcon in a few weeks. If you are interested in talking more about any of these 3 things or just want to chat about your vision of the future using CXL I am more than happy to chat!