dtaht

13 ms·

121.
by dtaht 3y ago
fq_codel was successfully offloaded in a few products, however, no offload here, just a xeon with a ton of cores and a big L3. On the libreqos case we leveraged some ebpf to do packet steering in what we call the heimdal bridge, and also ka
122.
by dtaht 3y ago
fq+ we have been running at 25+gbit or more, for quite some time. There are typically 32k FQ_codels in a 10gbit system. I am pretty sure, that we would not have scaled past 10gbit as well as we did, without having made fq_codel and sch_fq (
123.
by dtaht 3y ago
I hope more folk try this code out!!
124.
by dtaht 3y ago
Here is a realworld example, easily duplicated by anyone with a big fifo AND enough flows, that happens all the time, with cubic: https://blog.cerowrt.org/post/juniper/
125.
by dtaht 3y ago
This is a really good (and rare) interview with van, where he and mike finally asked the question - "how did this EVER work?": https://www.youtube.com/watch?v=QP4A6L7CEqA
126.
by dtaht 3y ago
exactly right. https://blog.cerowrt.org/post/juniper/ We have now produced variants of the flow queuing algorithm for fq-codel, fq-pie, and the FQ in cake, which we are now using in libreqos. It is the best form o
127.
by dtaht 3y ago
Dumb enough to me, includes basic fq and aqm technology at the router. FIFOs are the worst form of queuing. FQ solves most of the problem thoroughly enabling delay based transports, Van and Kathie´s codel, most of the rest. Van was also on
128.
by dtaht 3y ago
The article kind of missed that Van was the author of RED, and later he and kathie nichols did codel, and later he was on the team that did BBR. What would the internet have looked like without Van?
129.
Some proposed improvements to fq_codel and CAKE(docs.google.com)
2 points by dtaht 3y ago|1 comments
130.
by dtaht 3y ago
Over here I have been collecting feature requests and bug reports against the fq_codel and CAKE bufferbloat-fighting queuing disciplines, and (especially) seeking input from the field on problems observed and solutions found for things like
131.
by dtaht 3y ago
The embedded software industry in the US is nearly non-existent in the home router and IoT space. It is all leveraging ancient open source code, outsourced to the lowest bidder.
132.
by dtaht 3y ago
This is not really relevant to the proposed rulemaking but it is something that bugs me deeply, and would like to get off my chest. I would like to see a mandate that red LED be wired inband to every camera and microphone, on every device
133.
by dtaht 3y ago
This is a wonderful thread, and I am so glad to see so many sharing my nightmares and some of my conclusions. I encourage folk to work together on actually filing proposals with the FCC in their format. I would gladly join on such an effort
134.
by dtaht 3y ago
Thank you very much for reading. It was the first, and last time, I ever took place in the public process and political action. https://www.computerworld.com/article/2993112/vint-cerf-and-... We were within months
135.
by dtaht 3y ago
Agreed. I fully expect to see the routers we used in the cerowrt project from 2008 still operational for 10-20 more years. Thereś one out there with 4+ years of uptime that I know of. https://blog.cerowrt.org/post/an_
136.
by dtaht 3y ago
We have worked very hard in the OpenWrt and Linux projects to make it easy to update them in the field. Linux distros, android, apple, openwrt, etc have this facility built in now. IoT should also.
137.
by dtaht 3y ago
As these regulations change frequently, I am glad that linux makes it possible to update this database and the devices in the field. For example a portion of the 5.9ghz spectrum became available. https://www.computerworld.com&#x2
138.
by dtaht 3y ago
Vint Cerf and I shot that proposed anti-dd-wrt regulation down thoroughly: filing here: http://www.taht.net/~d/fcc_saner_software_practices.pdf Substitute IoT for router in everything we wrote there on page 12-13 and t
139.
by dtaht 3y ago
Thank you for engaging with the community in this way. Many years ago, in a fight to preserve individuals ability to flash their own routers, Vint Cerf, and I, and a coalition of many others, filed this report: http://www.taht.ne
140.
by dtaht 3y ago
Which TCP reserved bit? ECN? The reserved fields in the ack?
141.
by dtaht 3y ago
In general, none of the web based tests run for long enough. Some are configurable to run longer.
142.
by dtaht 3y ago
I have tried to produce bufferbloat-related tools that measure things far more accurately and in more detail. https://blog.cerowrt.org/post/flaws_in_flent/
143.
by dtaht 3y ago
yep. You can fix it for yourself with a SQM-enabled router... Or encourage your ISP to put in a better packet shaping solution in the first place. There are now a whole bunch of these - Preseem, Cambium, Paraqum, and my own libreqos. Cost i
144.
by dtaht 3y ago
For detailed bufferbloat related statistics, I helped create the flent.org tool, many years ago. It shows details that the web tests cannot. I keep hoping more folk will see networks with single digit latency under load stats, like what we
145.
by dtaht 3y ago
In some cases, the upload portion of the waveform test appears to be broken, btw. I worry that underneath it has switched from cubic to BBR and is thus understanding what a normal user actually experiences. Otherwise I still feel yours is t
146.
by dtaht 3y ago
Actually, while I like a well written corporate engineering blog, it is knowing that the individual voices within an org are allowed to blog, and for that matter, post emails to mailing lists, without oversight, that I find attractive. If a
147.
by dtaht 3y ago
I do not write corporate engineering blogs. I write truthful, and sometimes funny, engineering blogs. The blog used to be a source of genuine truth and engagement, now, they are too often sterilized, and boring. I know my writing would be
148.
by dtaht 3y ago
I use self hosted galene.org. It is FOSS & in go. Takes 10 minutes to setup. Is missing a few features.
149.
by dtaht 3y ago
That's what I do!
150.
by dtaht 3y ago
I lean towards a public benefit company for libreqos.io. I do wish more of the folk's business models we were actually helping (videoconferencing, gaming) would recognize their common interests with us, and chip in.

More ›