Discussion in Sec 5. brings up surprisingly little data

[larseggert]

Markku Kojo said:

I regret to say that the discussion in Sec 5. brings up surprisingly little data to back up the claims that are made. Given the long deployment experience that is emphasised in the draft, there, however, is little evidence (measurement data) summarised and cited to back up the claims. "There is a long deployment experience" does not provide any evidence as such. There should be a lot of studies with measurement data accumulated over the years that would support the assertions in the doc. Or, is there?

[larseggert]

@lisongxu, any update?

[larseggert]

@lisongxu ping

[sangtaeha]

@larseggert @lisongxu We have maintained a website for more than 10 years (2005 - 2016) showing the performance of CUBIC over many different networks. That site is gone now. I will try to find a comprehensive measurement study instead.

[sangtaeha]

@lisongxu We have a COMNET paper that evaluates CUBIC along with the other protocols w/ and w/o background traffic. Our INFOCOM convex ordering paper shows evidence of its stability. Another COMNET paper "Taming the elephant.." also has real-world results as well. Adding a few experimental study papers after that may be enough.

[larseggert]

If the site is on archive.org, we might be able to link to that?

[sangtaeha]

@larseggert Indeed, I found those. While I read Markku's comment about "There is a long deployment experience," I am not sure if there is better evidence than the fact CUBIC has been used in Linux for more than a decade and later Windows and macOS. While I am searching the measurement studies, there are too many papers to cite, which is not feasible, so I am adding a few references of mine and also a few recent studies.

@lisongxu There are a lot of measurement studies including ours and we may cite a few of them if needed.

• NS2 evaluation results, Dummynet testbed results, and internet evaluations
  https://web.archive.org/web/20161118125842/http://netsrv.csc.ncsu.edu/wiki/index.php/TCP_Testing

• Dummet experiments w/ and w/o background traffic
  https://web.archive.org/web/20161103143537/http://netsrv.csc.ncsu.edu/highspeed/

• CUBIC's stability experiments
  https://web.archive.org/web/20160122052654/http://netsrv.csc.ncsu.edu/convex-ordering/

• Technical Report on evaluating TCP congestion control protocols
  https://netstech.org/asteppaper.htm

• Internet evaluation of HyStart
  Sangtae Ha and Injong Rhee, “Taming the Elephants: New TCP Slow Start,” Elsevier Computer Networks, vol. 55, no. 9, pp. 2092–2110, 2011.

• Impact of background traffic to CUBIC and the other TCP congestion control protocols.
  Sangtae Ha, Long Le, Injong Rhee, and Lisong Xu, “Impact of Background Trac on Performance of High-Speed TCP Variant Protocols,” Elsevier Computer Networks, vol. 15, no. 4, pp. 852–865, 2007.

• Stability analysis of CUBIC
  Han Cai, Do Young Eun, Sangtae Ha, Injong Rhee, and Lisong Xu, “Stochastic Convex Ordering for Multiplicative Decrease Internet Congestion Control,” Elsevier Computer Networks, vol. 53, no. 3, pp. 365–381, 2009.

• Two recent measurement papers among many
  Salman Muhammad, Touseef Javed Chaudhery, Youngtae Noh, "Study on Performance of AQM schemes over TCP Variants in Different Network Environments,", IET Communications, Vol. 1(19), 2021.
  J. Gomez, E. Kfoury, J. Crichigno, E. Bou-Harb and G. Srivastava, "A Performance Evaluation of TCP BBRv2 Alpha," 2020 43rd International Conference on Telecommunications and Signal Processing (TSP), 2020, pp. 309-312

[larseggert]

Let's add some text about and references to some (or all) of those and assume the issue is addressed.

[larseggert]

@lisongxu/@sangtaeha could we get a PR for this, please?

[lisongxu]

@larseggert I will get a PR this weekend. Thanks