A paragraph about networks and queue theory?

[dtaht]

I was so glad to see you cite queue theory in your readme. I have written in many places about queue theory in general, as to how it applies to networks.

In particular, a treatment of "little's law" needs to done with great caution, as shedding load on networks is different than balancing load for disks, and I've always longed for more folk to have read Kathie Nichols & Van Jacobson's great paper on the codel algorithm here: https://queue.acm.org/detail.cfm?id=2209336 - or read up on fq_codel, which is usually the default network queuing algorithm in Linux nowadays.

Over here, last year, was a rollicking debate between Len Kleinrock (Father of modern queue theory) and Dave Reed (author of the end to end argument and creator of UDP) about how to think about little's law on networks: https://lists.bufferbloat.net/pipermail/cake/2021-July/005540.html

especially about the flaws in little's law - yet several folk in that discussion held on doggedly to their belief system. As much as I think more and more knowledge of how queues work is propagating to the SRE's and as much as I love flame graphs... in terms of trying to add a paragraph onto your stuff, I find that I do not know what y'all don't already understand?

PS: I only discovered your flamegraph tool today (I'd been using greggs work for years). It's awesome!

[djc]

Hey, glad you're liking this tool! I'm a replacement maintainer for this tool, and am much less experienced/opinionated in terms of those sections of the README than the original maintainer (@spacejam) -- after all, in the end this tool just calls out to perf (or other supported sampling profilers on other platforms) and the inferno Rust library, so in that sense it doesn't really require in-depth knowledge to maintain.

I'm not also not super clear on what you're suggesting/requesting with this issue, but if you have some useful stuff to add, please feel free to submit a PR.