RIPE 67

Minutes of the Open Source Working Group

Date: 16 October 2013, 09:00-10:30

Co-Chairs: Martin Winter and Ondrej Filip

Scribe: Daniel Quinn

Status: DRAFT

A. Administrative Matters

Ondrej Filip, Open Source Working Group co-Chair, welcomed everyone to the session, which he called a “historical moment” because it was the first session of the Open Source Working Group.

B. Overcoming traditional network limitations with open source

Nat Morris, Cumulus Networks Inc.

This presentation is available at:

https://ripe67.ripe.net/presentations/235-Cumulus_Networks_-_Overcoming_traditional_network_limitations_with_open_source.pdf

Stéphane Bortzmeyer, AFNIC, suggested that the speaker was ignoring an alternative explored at the IETF ForCES working group. He asked if Nat had heard of it or had any opinion about it.

Nat responded that the interest was in reducing costs, and that they chose to tackle this problem by targeting existing hardware rather than looking into the project Stéphane  mentioned.

Kostas Zorbadelos, OTE SA, said he was interested in how hardware acceleration was achieved. He asked if Nat’s company produced the kernel module, what his business model was, and if the module was open source.

Nat confirmed that the module was GPL and explained that the Linux distribution was just Debian Weezy. He said the business model was that

people can negotiate their own prices through ODMs or resellers, and then install Cumulus software on the hardware.

Kostas asked if the GPL is an obstacle.

Nat replied that all the changes made have been released and that they were currently focused on Broadcom devices.

Kostas noted that Broadcom does not have a good reputation for open source development.

Nat went on to explain that the model they were developing focused their business on the HAL layer, to which Kostas acknowledged that, “the magic

is in the HAL layer”.

Michael Abrahamsson, T-Systems, asked if the reason this was possible was that switches have improved in terms of hardware in recent years.

Nat confirmed this, and went on to explain that in more marketing-related presentations, this relationship was deliberately stated.

Michael inquired as to the problems presented by binary drivers. Nat noted that elements of the SDK are binary, but said he was hopeful that this will change in time with efforts like this project.

There were no further questions.

C. Building and Maintaining the Nagios Ecosystem

Scott Wilkerson, Nagios

The presentation is available at:

https://ripe67.ripe.net/presentations/200-Scott_Wilkerson_-_Building_and_Maintaining_the_Nagios_Ecosystem.pdf  

Martin Winter, Open Source Working Group co-Chair, mentioned that it’s great that Nagios is around and supported for so long.

Scott pointed out that while they’re a small company with limited funds, they are not a non-profit and they are profitable.

Martin asked what the difference was between Nagios Core and Nagios XI.

Scott explained that XI is built on Core, but contains a series of other open source projects. It includes handy configuration scripts and wizards. XI has a user and session-based authentication, per-user custom notifications, advanced reporting features with extrapolation, an extendable interface in PHP, and is potentially more user friendly.

Bengt Gördén, Resilans AB, wanted to know if the commercial vendors actually contribute back to the community.

Scott said his experience was that many of the vendors do contribute back, both in the form of patches and development on Core.

There were no further questions.

D. NSD 4.0 Update

Willem Toorop, NLnet Labs

The presentation is available at:

https://ripe67.ripe.net/presentations/232-nsd4.pdf

Shane Kerr, Internet Systems Consortium (ISC), asked why he used libevent, and if other models for using different sockets were explored. Specifically he asked about “libev”.

Willem responded that he didn’t know.

Ondrej wanted to know the versions used for the graphs in the presentation. Willem responded that the version of Knot was 1.2.0. Ondrej commented that that was a very old version.

Jaap Akkerhuis, NLnet Labs, attempted to answer Shane’s question, stating that libevent was the most portable, and that’s why it was selected.

There were no futher questions.

E. What to Expect from Open Source WG & Closing Remarks

Martin encouraged everyone to join the Open Source Working Group Mailing List and get involved. The floor was then opened to ideas and questions directed toward the Open Source Working Group in general.

Michael Abrahamsson, T-Systems, wanted to talk about how you combine the two needs of rapid, fragmented development and the need for strong, stable software.

Carsten Strotmann, Men & Mice, wanted to hear about use cases for various software projects and what works and doesn’t.

Martin suggested that the working group cover smaller pet projects that work but don’t get enough attention. He also suggested that for these smaller projects, the working group encourage those people to show up to the RIPE Meetings.

Martin thanked everyone for attending and closed the session.