When I was at the nznog conference there was a lot of despair about a lot of aspects of the fiber deployment, and the quality of the home gateways, and what seemed to be promising plans to try to get openwrt with fq_codel running on nzwrt... did you all despair completely? I talked about all the great things we could do to improve our devices with openwrt, and and beat bufferbloat in particular with fq_codel in my talk... and since then I have worked to get fq_codel in more and more OSes and hardware. UBNT has added configuration tools for it in their upcoming edgerouter 1.7 release. VyOS just added it. It is in click. It is in fedora 22. It is spreading (And of course, already in all modern linuxes, openwrt, dd-wrt, ipfire, shorewall, etc). There was no sign of a pulse at tilera or mikrotik, but I will keep trying. Still... to one day get hardware that has low latency and good queue management at gigE speeds or higher requires better hardware offloading techniques than the bulk notions like GRO and TSO that exist today. And it has long been obvious that the rate limiting + fq_codel techniques we use successfully at sub 200mbit speeds do not scale higher. So the algorithms need to move into hardware, also. *******so... a brief commercial interruption***** I think the open source design method has worked spectacularly for software, and I think the same methods can work for hardware if only we can pull together larger communities to do the work, worldwide. There are only 24 hours left on this kickstarter - we CAN *start* to take back the edge of the internet - if we can only find another 5k of funding. https://www.kickstarter.com/projects/onetswitch/onetswitch-open-source-hardw... This FPGA board´s pcie interface and switch design - and the split memory interfaces between the programmable logic and the onboard dual A9 core - and reducing the cost from 7000 to 700 bucks - are the important parts to why the community needs this board - in my case - so that more of htb + fq_codel can move into hardware that anyone can build and use - in other cases? who knows what could be done with it!? One day, we could move the logic into asics, and finally have open source, blob-free hardware to work with for any network purpose. There are people on these lists with money, and there are those with talent and time, and it would be great if more of those people could line up with each other. I put in all I could spare (8500 dollars) into this kickstarter. I have one of their high end boards, already. It´s great. There is a get one give one program that I asked meshsr to put in to try and enable that event. If there is anyone here that would like to help hack together the next generation of edge network hardware, after this kickstarter completes, let me know. I return you now to your normal despair about vendors not listening, and crappy OSes you can´t otherwise fix. -- Dave Täht We CAN make better hardware, ourselves, beat bufferbloat, and take back control of the edge of the internet! If we work together, on making it: https://www.kickstarter.com/projects/onetswitch/onetswitch-open-source-hardw...