Open Source Working Group Minutes RIPE 78

Session: Wednesday, 22 May, 14:00 – 15:30
Chairs: Ondrej Filip, Martin Winter
Scribe: Anand Buddhdev
Status: Draft

A. Administrative Matters

The presentation is available at:
https://ripe78.ripe.net/presentations/92-RIPE78-Opensource-WG-Agenda.pdf

Ondrej Filip welcomed everyone to the session. Martin Winter ran through the agenda. Ondrej Filip declared the minutes from RIPE 77 approved.

B. OpenSource Based Router Labs

Wolfgang Tremmel, DE-CIX

The presentation is available at:
https://ripe78.ripe.net/presentations/57-57-2019-05-22-RIPE78-MANRS-Lab.pdf
https://ripe78.ripe.net/presentations/74-74-e-wt-20190123-UsingDockerForTeaching-01.pdf

Owen DeLong from Internet NZ asked about what kind of hardware one should use for a 20-person lab.

Sander said he ran on Dell R710 with 32GB of RAM and 8 CPU cores, and it was struggling with 30 students. So this lab would need more CPUs and memory.

Ruben van Staveren from Proserve commented that this work looked interesting and wanted to try it out.

C. High-Performance Traffic Encryption on x86_64

Max Rottenkolber

The presentation is available at:
https://ripe78.ripe.net/presentations/13-high-performance-traffic-encryption-on-x86.pdf

Tom (Cloudflare) said it was a cool talk and a very good project. He asked whether the numbers were with Intel exploit mitigations turned on or off in the kernel. Max said he didn't think it would make a difference, because it's not using syscalls or kernel.

Jordi Palet asked whether Max had considered doing something similar in low-cost chipsets such as those used in CPEs, for example, OpenWRT. Max replied that at the moment they only support x86_64 because they need some specific features. He wasn't sure if it would be possible to use other chipsets, but they could probably run a low-cost box for up to 1Gb/s of traffic.

Blake Willis asked how many networks were using this in production. Max said none, because this is a prototype.

Brian Dickson (GoDaddy) asked about the state of drivers for NICs. Max said it works now with the Intel 10G NICs and Mellanox. He is working on support for Intel AVF. He said that drivers were available for Connect X cards. Brian asked about hyper-threading and Max said one should enable and use hyper-threading, because one gets better performance.

D. Open Source Lightning Talk(s)

The presentation is available at: 
https://ripe78.ripe.net/presentations/91-bird_new_features_migrate_ripe78.pdf

Sander Stefan made 2 small requests:

1. He presented the NAT64 check website at the previous meeting, and asked for help with maintaining the website; and

2. He called for volunteers for development help with the global NOG alliance.

Maria Jan Matejka from CZNIC presented an update about BIRD version 2.

There were no comments or questions.

E. Roundtable Discussion: The role of open source in industry hackathons and how to best serve their communities

Charles Eckel, Cisco and Mirjam Kühne, RIPE NCC

The presentation is available at:
https://ripe78.ripe.net/presentations/99-RIPE78-OpenSource-Hackathons-Eckel-Kuehne.pdf

Spencer Dawkins said that the hackathons at the IETF made a big difference in producing running code and protocol development quickly. Mirjam asked whether there was a mailing list, and Charles said that there was hackathons mailing list.

Tim Wattenberg said he has been to two IETF hackathons, and found them very useful, because authors of drafts are often there, and can answer questions. He highly recommended the hackathons.

Randy Bush said he's been to many hackathons. He said that there was very little cross-fertilisation. There was not much interaction with other tables. He found that the RIPE hackathons were more researcher- and operator-friendly, compared to the IETF ones. He said that the IETF hackathons are too big and congested. In comparison, the RIPE hackathons were roomier and allow more space.

Spencer Dawkins said that since the last IETF, there have been coding lounges available.

Benno Overeinder said he understood Randy, but felt that the IETF hackathons were very useful, and perhaps fit with the IETF's goals.

Alyssa Cooper wanted like to know how to make IETF hackathons more attractive to operators.

Mirjam said that Charles had written a blog about IETF hackathons, and it has also been published on RIPE Labs. She said that on RIPE Labs, there were also posts about previous RIPE hackathons.

Martin Winter liked the IETF and RIPE hackathons, but has been to others which are not so well organised. He asked whether Mirjam and Tom could provide more info for hackathon organisers to make their hackathons better. Mirjam said that Vesna Manojlovic from the RIPE has a lot of experience and documentation that RIPE NCC can make available.