does someone else see a need for running VyOS on standard network device hardware like switches etc?
Switching and routing gets more and more integrated and in modern datacenters you rely on solutions like EVPN-VXLAN when you want to be independent from a single manufacturer.
Until Nvidia has bought Cumulus Networks there was a player which provided a solution on white label switches.
Now it seems like only the hyperscalers are making use of existing open source platforms like Open Network Linux.
Personally I really would like to run VyOS also on switch hardware especially on smaller ones which are currently missing such features.
Eventually this could be possible by combining VyOS and Open Network Linux.
I think running VyOS on EdgeCore Switches could be super. We would „only“ need to add switching support (support the ASIC with the Linux Switching drivers).
That wont be different then Arista EOS which also run Linux (I think Fedora?) under the hood of their CLI. There are not that much Switching ASICs out there.
We need to center on switchdev support. so no need for any additional SDK (that is nearly impossible to get access to).
Personally, I’m interested in 10g switches for non-data center applications
YES to this! I haven’t yet heard of switchdev until @syncer mentioned it, but it definitely looks promising. We can currently use bridging but need to leverage the raw performance of ASICs in most cases. Being able to have a more consistent route/switch platform (as IOS somewhat provides) is very compelling.
Isn’t this essentially the same idea as the DANOS project? Given the common(ish) heritage of Vyos and DANOS, I wonder if there is an opportunity to work towards merging the 2 projects rather than competing?
I haven’t yet heard of switchdev until