Child pages
  • Teleconference notes September 25, 2014
Skip to end of metadata
Go to start of metadata

-----------
wg-SDN call notes, 9/25/2014

Attending:
kevin mayeshiro, kmayeshiro@ucdavis.edu
Michael Lambert <lambert@psc.edu>
Dan Schmiedt, willys@clemson.edu
Matt Zekauskas, matt@internet2 for the first 30 min
Kathy Benninger, benninger@psc.edu
{also heard on the call: Paul Schopis, Eric Boyd Chris ? from Clemson, Ken Miller (PSU)
Roy Hockett, royboy@umich.edu

0) Agenda Bash

AJ: new FlowSpace Firewall release.  Fixes a bug, should move to 1.0.3-2
AJ: Have a release candidate for SCIpass (which I believe can be characterized as a way to have science flows automatically bypass firewalls --mjz)

1) AL2S roundtable: how are you connecting to AL2S and what are you doing with it?

Eric Boyd: Multi-domain capability is a focus.  Planning to demo network slicing across multiple admin domains at TechX.

Chris Konger: Clemson is using L3 and L2.  Static routes for L3.  L2 connections as well.  Connecting infrastructure is brocade MLXe32. GENI VLANs are extended to iON in ATL.  Turning off MAC learning for OF can be challenging.  Various (mostly hardware/switch related) challenges related to internal OF-based science DMZ.  PerfSonar is used for testing/throughput testing.  Have servers facing legacy inet and others facing AL2S.

Paul Schopis: Much work transferring medical images.  Looking at turning up new (2nd) 100G to Cincinnati.

Akbar Kara: multiple 100G and 10G interfaces to AL2S.  Use for GENI, with multiple protected paths.  Upcoming connectivity to IBM for bluejean research work.

Michael Lambert: 100G link mostly for L3.  First stages of testing for CC-IIE project.  Static links, not much dynamic SDN yet. 

Eric Boyd: what would change in the way you'll do things in the future?  ML: static configuration is fine for the foreseeable future.  Concern about stability of protocols for dynamic circuits.  Parallels to dynamic ATM circuits.

Chris Konger: Pain points involving connectivity to STARlight, etc.

[dan leaving call]

  • No labels