Quote:
Originally Posted by dragon
If you wanted to bond at those speeds I believe it's possible using off the shelf X86 hardware and something like Vyos or RouterOS, that said the required config often is rather complicated particularly if it's unbalanced lines and you need to skew the ratios to take that into account.
|
pfSense does it too, and of course if you're a simpleton you can run dd-wrt or Openwrt with their 1-click bonding GUI on x86 as well. If you're a complex-ton, like a masochistic challenge, or just have a lot of manpower and expertise to show off, you can run a custom distribution of RHEL on a cluster of rack servers with an LCFG automated configuration backend for your routing infrastructure...
Any old consumer router will support wire-speed forwarding with VLANs so a cheap old spare laptop somewhere will do the trick. Some laptops can provide a wireless AP as well all while using less power than the Superhub!
TBH unbalanced
bonding is easy. You specify the upstream and downstream rates on each connection and you're done. What's complicated is dynamically dealing with VM's STM modifying your upstream rate and having something at the other end to terminate your aggregated link.
I suspect you may actually mean load-balancing/failover, which is indeed awkward to balance when you have unbalanced lines, latencies, and routes.