TERENA/DANTE TASK FORCE FOR TESTING ADVANCED NETWORKING TECHNOLOGIES

Minutes of the 5th TF-TANT meeting held on the 30th of September 1999 at
the Steigenberger Airport Hotel, Frankfurt, Germany.

Kevin Meynell - Issue 2

PRESENT

Name                        Organisation        Country
----                        ------------        -------
Luca Dell'Agnello           INFN/GARR           Italy
Hamad el Allali             U.Twente            Netherlands
Claudio Allochio            GARR                Italy
Werner Almesberger          EPFL                Switzerland
Kurt Bauer                  U.Vienna/ACOnet     Austria
Spiros Bolis                GRNET               Greece
Massimo Carboni             GARR                Italy
Zlatica Cekro               VUB/ULB             Belgium
Phil Chimento               U.Twente            The Netherlands
Larry Dunn                  Cisco               United States
John Dyer                   TERENA              -
Hans Joachim Einsiedler     T-Nova Berkom       Germany
Tiziana Ferrari             INFN Bologna        Italy
Iphigenie Foumta            GRNET               Greece
Ruediger Geib               T-Nova Berkom       Germany
Silvia Giordano             EPFL                Switzerland
Leon Gommans                U.Utrecht/Cabletron The Netherlands
Christoph Graf (Chair)      DANTE               -
Avgust Jauk                 ARNES               Slovenia
Joop Joosten                CERN                Switzerland
Dimitrios Kalogeros         GRNET               Greece
Tom Kosar                   CESNET              Czech Republic
Olav Kvittem                Uninett             Norway
Cees de Laat                U.Utrecht           The Netherlands
Simon Leinen                SWITCH              Switzerland
Lladislav Lhotka            CESNET              Czech Republic
Dimitrios Matsakis          GRNET               Greece
Kevin Meynell (Sec)         TERENA              -
Mike Norris                 HEAnet              Ireland
Jan Novak                   DANTE               -
Simon Nybroe                Ericsson Telebit    Denmark
Herman Pals                 KPN Research        The Netherlands
Yiannos Pitas               CYNET               Cyprus
Alex van der Plas           Ericsson Telebit    Denmark
Esther Robles               RedIRIS             Spain
Roberto Sabatino            DANTE               -
Trond Skjesol               Uninett             Norway
Robert Stoy                 RUS/DFN             Germany
Celestino Tomas             RedIRIS             Spain
Panagiotis Tzounakis        GRNET               Greece
Jean-Marc Uze               RENATER             France
Guido Wessendorf            U.Muenster/DFN      Germany
Bert Wijnen                 IBM                 The Netherlands
Wilfried Woeber             ACOnet              Austria

Apologies were received from:

Michael Behringer           Cisco               United Kingdom
Juergen Rauschenbach        DFN                 Germany
Victor Reijs                SURFnet             The Netherlands


1.  APPROVAL OF MINUTES

     The minutes of the TF-TANT meeting held on the 17th and 18th of
     June 1999 were approved.


2.  STATUS OF QUANTUM & TEN-155

     Roberto reported that KPN would be upgrading a number of circuits as
     compensation for late delivery of other circuits. An STM-1 from
     Italy to Switzerland would be delivered on the 1st of October, a T3
     from Belgium to France would be delivered on the 15th of October,
     whilst an E3 from Spain to Switzerland would be delivered on the
     15th of December. The latter connection would share the same
     infrastructure as Spain to France.

     In May, an Invitation for Tender was issued for an interconnection
     agreement. This called for connectivity to commercial networks in at
     least three locations, and would replace existing peering
     agreements. Replies were received from AUCS, BT, Ebone, IBM, KPNQ
     and UUNET.

     The evaluation criteria was based on the number of available
     exchange points, the cost of the local loops to TEN-155, transit to
     all major ISPs, and usage of symmetric routing. As a result of this,
     AUCS (ATT Unisource Commercial Services) were selected, and from the
     1st of October, interconnections will be established at 45 Mbps in
     Amsterdam and 25 Mbps in Switzerland. At a later date, additional
     interconnections at 20 Mbps will be provided in London and
     Stockholm, whilst Amsterdam will be downgraded to 15 Mbps. The
     contract also specifies that AUCS must peer with any unreachable ISP
     when traffic to that destination over other routes reaches a certain
     volume.

     An MoU had been signed with UCAID to allow TEN-155 to peer with
     Abilene. A 45 Mbps line had already been procured to connect the
     DANTE and Abilene routers in New York. This would enable all
     countries currently using the TEN-155 US service to use the
     Internet2 backbone.

     European Commission funding for TEN-155 was due to finish on the
     31st of May 2000, but the new Head of DG-XIII/G2 (? Campolargo) had
     verbally assured DANTE that funding would continue until the start
     of the Fifth Framework Programme. With regard to this, a Call for
     Proposals had recently been issued with two areas related to
     research networking: RN1 - interconnection of NRNs, and RN2 -
     testbed networks. Proposals were due at the beginning of next year.

     Avgust enquired how the AUCS peering agreement would be monitored.
     Roberto replied that DANTE would provide traffic statistics to and
     from AUCS, but the monitoring of traffic to ISPs not covered by the
     agreement would be the responsibility of the NRNs.

     Tiziana asked whether a Managed Bandwidth Service was available to
     Abilene. Roberto replied this was theoretically possible as the
     connection to New York ran over ATM. Nevertheless, Abilene itself
     did not use ATM, so service could not be guaranteed over this
     network (unless of course IP QoS was introduced).

     Jean-Marc asked about the rationale for using ATM to Abilene.
     Roberto replied an ATM connection was the best offer they had
     received.


3.  QBONE PRESENTATION

     Ruediger gave a presentation on the Internet2 QBone initiative
     (http://www.internet2.edu/qos/qbone/).

     In recent months, work had been progressing on producing the
     necessary specifications for the QBone. The first versions of the
     Premium Service specification which described an inter-domain EF-
     based service, and the interoperability test description were now
     available. The bandwidth broker specification was still being
     developed.

     A number of institutions were currently working on bandwidth
     brokers; British Columbia Institute of Technology, GLOBUS, MERIT,
     Telia/University of Lulea and UCLA. Two phases had currently been
     defined. Local admission control and traffic conditioning would be
     initially developed, with inter-bandwidth broker signalling to
     follow.

     An operability workshop for bandwidth brokers was due to be held in
     November. The functionality of different implementations would be
     compared so improvements could be identified. At the request of
     the vendors, this event would not be public.

     A QCon Interoperability Workshop would also be held from 1-5
     November 1999 (http://www.cnl.ncsu.edu/interop/diffserv/). This was
     being hosted by NCNI, and would feature six vendors. It aimed to
     validate the conformance of bandwidth brokers to RFCs 2475 & 2598,
     and the Protocol Implementation Conformance Statement (PICS). This
     event was also closed to the public, but it was hoped future events
     would be more open.

     Christoph thought the QBone activities seemed rather secretive and
     wondered how TF-TANT could contribute. Ruediger replied anyone could
     cooperate with the QBone if they were willing to adopt the defined
     model. Most of the active development of bandwidth brokers was in
     Europe anyway, and collaboration would be beneficial to all parties.

     Tiziana however, believed TF-TANT should concentrate on its own test
     programme for now, but could perhaps collaborate with QBone in the
     new year.

     Tiziana also asked why an AF-based service had not been specified.
     Ruediger replied it was important to keep the QBone simple in the
     initial phase.


4.  EXPERIMENT DISCUSSIONS

     4.1 Flow-based Monitoring

     Simon reported that traffic flow data was now being collected at the
     TEN-155 PoP in Switzerland. This currently only utilised 2-3% of the
     CPU in the router, but there were still problems collecting
     multicast statistics. The aim was to evaluate the data with
     different analysis tools in order to determine which produced the
     best results, and to provide feedback to developers.

     Tom said he would be willing to grant access to the CESNET TEN-155
     routers for monitoring purposes.

     Dimitrios mentioned a flow replicator that had been written at GARR.
     Unfortunately, no-one appeared to know who developed it.

     Roberto believed there should be some investigation into the
     accuracy of different sampling rates. It was currently possible to
     analyse every packet on a 155 Mbps network, but this was unlikely to
     scale as line speeds increased.

     4.2 IP Version 6

     Wilfried gave a presentation on the IPv6 network at the University
     of Vienna (URL?). This generally worked well, but as it was being
     used to test IPv4 and IPv6 interoperability, it was unclear whether
     IPv6 hosts were communicating with the DNS via IPv6. There were also
     issues relating to the default routes for both IPv4 and IPv6.

     Guido gave a presentation on JOIN (http://www.join.uni-muenster.
     de/). This was a German project to establish both an IPv6
     interoperability testbed, and a native IPv6 network over the B-WiN
     network.

     Bert asked whether anyone had investigated renumbering. Wilfried
     replied they were interested in renumbering, but DNS and
     particularly reverse DNS implementations were not yet available.
     These would be necessary before a pilot service could be offered.

     Wilfried requested input from DANTE about what should appear in his
     final report.

     ACTION 5.1 - DANTE

     4.3 Differentiated Services

     Tiziana reported this activity had started at the end of June and
     currently involved nine sites. There were some problems with the GPS
     antennas on the SmartBits equipment, but NTP could be used if
     absolutely necessary. Netcom had also agreed to extend the loan
     period. The next stage was to increase the number of nodes, and test
     equipment from other vendors.

     Tiziana suggested the interim test results could be documented in an
     Internet draft for IETF46. This would need to be submitted before
     the 22nd of October 1999, and she asked participants for input.

     ACTION 5.2 - DiffServ Participants


5.  REVIEW OF EXPERIMENTS

     5.1 RSVP to ATM Mapping

     Tiziana said this activity had been constrained by the availability
     of SVC signalling on TEN-155. Nevertheless, tests could now
     commence.

     Larry mentioned that Laura Cunningham (lcunning@mci.net) and John
     Jamison (jjamison@mci.net) had a lot of experience in this area and
     it might be an idea to contact them for advice.

     5.2 MPLS

     Jean-Marc said an interim report was now available on the Web
     (URL?). Unfortunately, a bug in the Cisco architecture had meant
     some of the tests could not be completed. This had been reported,
     but no solution was forthcoming as yet.

     The next phase would be to test MPLS with QoS, and to conduct
     interoperability tests with other implementations such as Juniper,
     Nortel Networks and Telebit. This would probably take place at the
     beginning of next year.

     5.3 Multicasting (IP and ATM)

     Robert reported that MSDP/MBGP was now running on TEN-155 between
     France, Netherlands, Sweden and the UK (http://www.dante.net/
     mbone/). The next stage was to migrate the whole of TEN-155 to MBGP.

     The MECCANO multicast network over the MBS had also improved
     dramatically since MBGP had been introduced. There had been some
     problems initially, but these had largely been resolved. The project
     might now consider using 'production' multicast services, provided
     they could reliably obtain 5-10 streams of 500 Kbps.

     In the future, MASC/MBGP should be tested now there was an
     implementation from ISI. In addition, some aspects of BGMP had been
     integrated in gated.

     Christoph asked whether ATM point-to-multipoint had been tested. Jan
     replied there had not been any progress as the tests were dependent
     on signalling, and SVCs were only available with CBR. It might be
     possible however, to conduct some tests in a controlled environment.

     5.4 ATM Signalling

     Jan reported that signalling (using CBR) was now available on TEN-
     155, although further tests on connecting NRNs via SVCs still had to
     be completed. Unfortunately, no NRN currently planned to offer its
     own signalling service, so experiments would be limited to the
     backbone.

     Christoph said it was unclear whether anyone actually required
     signalling nowadays. It was considered essential one or two years
     ago, but lack of availability meant other solutions were now being
     considered. Wilfried added that unless it could be demonstrated that
     signalling provided real advantages, it should not be pursued
     further.

     Cees believed a major drawback of signalling was the lack of policy
     management. Until this was available, there would be few compelling
     reasons to use it.

     Dimitrios asked whether signalling could be used to configure
     routing on TEN-155. Roberto replied this would be investigated, but
     it would be difficult to ensure load balancing and circuit
     restoration.

     5.5 Policy Control

     Leon reported he was currently investigating policy control for
     DiffServ. Policy control servers were being developed by Cisco, IBM
     and Cabletron, and he hoped some trials could be run during the
     first quarter of next year.

     Tiziana (IBM), Philip (Cisco) and Cees (Cabletron) expressed
     interest in participating in these trials.

     5.6 Route Monitoring

     Simon L reported there had been no progress since the last meeting.
     Some work was being done for DANTE, but not in conjunction with this
     activity. It would also be a bit problematic to set-up the testing.


6.  EXPERIMENT SCHEDULE

     Christoph summarised the experiment schedule:

     The IPv6 and DiffServ activities were currently competing for
     network resources so SWITCH and RedIRIS were experiencing problems.
     They could however, use tunnelling if necessary.

     It had been agreed that a permanent overlay test network on the MBS
     was not particularly useful. This would not be established.

     There would be no further MPLS activity until early next year.

     It was too early to establish a QoS pilot service. This might be
     possible in around six months.

     The flow-based monitoring, policy control and multicast activities
     were ongoing.


7.  DATE OF NEXT MEETING

     The next meeting will be held on the 25th and 26th of November 1999
     at RedIRIS, Madrid, Spain.


8.  ANY OTHER BUSINESS

     Kevin mentioned that APAN (Asia-Pacific Advanced Network), and
     SingAREN (Singapore) in particular, were interested in collaborating
     with TF-TANT. Common areas of interest were identified as: IPv6,
     Multicasting, DiffServ and Policy Control. Some institutions were
     interested in a joint testing programme if suitable connectivity
     could be arranged.

     Tiziana suggested the DiffServ participants hold a meeting during
     the Washington IETF, as most people would be going.

     Christoph announced he was standing down as the Task Force Leader
     due to increased commitments at SWITCH. He proposed Roberto as his
     successor, and this was unanimously approved by the group.

     Cees said he planned to bring a wireless network hub to the next
     meeting. This used the same technology as the IETF meetings, so
     anyone with a laptop and appropriate PCMCIA card would be able to
     interwork.


9.  ACTIONS FROM LAST MEETING

     4.1  Kevin Meynell to contact Advanced Networks about the
          possibility of loaning some Surveyor equipment.
          - Superseded. It was decided to use the SmartBits equipment
          instead.

     4.2  Jean-Marc Uze to approach Netcom about the possibility of
          loaning more SmartBits equipment.
          - Done.

     4.3  All IPv6 experiment participants to supply information about
          their available equipment, bandwidth and manpower.
          - Ongoing. Only SURFnet had supplied this information.

     4.4  DANTE to create a mailing list for the IPv6 experiment.
          - Done.

     4.5  Jan Novak to include the KPN addressing plan in the ATM
          signalling experiment proposal.
          - Done.

     3.3  Tiziana Ferrari to incorporate bandwidth brokers into the
          Differentiated Services experiment proposal.
          - Done.

     3.8  Victor Reijs to draft document expressing the concerns of the
          research community about STM-4c.
          - Status unknown. Victor was not at the meeting.


OPEN ACTIONS

     5.1  DANTE to provide input to IPv6 report.

     5.2  DiffServ participants to provide input to Internet draft by
          22/10/99

     4.3  All IPv6 experiment participants to supply information about
          their available equipment, bandwidth and manpower.

     3.8  Victor Reijs to draft document expressing the concerns of the
          research community about STM-4c.