Lastest Architecture Minutes - Artur

Attending: Roberto Sabatino, Tony Barber, Erik-Jan Bos, Bruno Hoeft, Ronald van der Pol, David Foster, Dale Finkelson, Mike O'Connor, Azher Mughal, Shawn McKee, John Bigrow, Bill Johston, Ryan Harden, Ian Gable, Roy Hockett, Joe Mambretti, Michael Ernst, Artur Barczyk

Bill is collecting input and updates to the connectivity diagrams.
Please forward comments to him directly.
 
Bill: a policy document is needed. What's expected when an institute
joins LHCONE. Including Technical connectivity, security, AUP.
Bruno: this should be a shared document (all agree)
A discussion item will be added to the agenda on Friday (done).
 
Dale: what is the perception from the user community of the
LHCONE so far?
Michael: US Atlas is following the connectivity matrix, the change
is positive.
Shawn: E.g. when UMICH joined LHCONE, the result was immediate
throughput increase to/from European Tier2s. (kB/s to 10s MB/s)
 
Shawn: Move of AGLT2 to LHCONE caused non-Tier2 related IPs to loose
connectivity to CERN.
Ryan: that's the known issue with asymmetric routing and stateful
firewalls. Appears where PBR is not possible. Sites need
to be made aware of this kind of potential issues - we need to
document the gotchas, too.
Mike: his presentation from the last meeting contained much of
such information, could be used as first pointer, but needs text
put around.
 
Due to many of us travelling next Monday, we'll drop the call,
and continue the discussion directly at the Oslo meeting.
(With E-mail as communication channel in between.)
 

Submitted by David Foster on Tue, 09/11/2012 - 09:23