Where is this alleged bufferbloat coming from?
It can’t be from rate queues. The highest we set our Mikrotik queues is around 40 packets before they start dropping packets. We have pushed the queue depth higher to signal congestion to TCP Vegas style implementations. But at 10 Mbps that’s still only ~40 milliseconds of delay. I don’t think that qualifies as bufferbloat. Where in a typical WISP network are these huge buffers? Are you talking about APs at 100% of capacity? I admit I don’t know how much data an AP will buffer waiting for a timeslot to send the data over the air. But the only time I see latencies soar toward 1 second under load is on my one hated WiMAX basestation, and I think that may be due to excessive HARQ retries or something. From: AF <af-boun...@af.afmug.com> On Behalf Of Dev Sent: Tuesday, November 13, 2018 11:41 AM To: AnimalFarm Microwave Users Group <af@af.afmug.com> Subject: Re: [AFMUG] bandwidth management appliance opinions I looked at a couple variations of buffer bloat management, and have decided to build my own and maybe just open source the thing for “people who feel 50K seems excessive” and just need some basic functionality on a vanilla Linux box. The open source tech is out there, it’s just tying it all together in some sane way. I hope others will open source what they’re working on too, that’s what the community is about. I feel like the community is moving away from including the little guys these days. <http://af.afmug.com/mailman/listinfo/af_af.afmug.com> http://af.afmug.com/mailman/listinfo/af_af.afmug.com -- AF mailing list AF@af.afmug.com <mailto:AF@af.afmug.com> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
-- AF mailing list AF@af.afmug.com http://af.afmug.com/mailman/listinfo/af_af.afmug.com
