Minutes of WG-LLT meeting, Dresden, 4/10/1998
|
|
|
TERENA WORKING GROUP ON LOWER LAYER TECHNOLOGIES
Minutes of the WG-LLT meeting held on the 4th October 1998 at the Hotel
Mercure Newa, Dresden, Germany.
Kevin Meynell - Issue 3
PRESENT
Name Organisation Country
---- ------------ -------
Zlatica Cekro VUB/ULB Belgium
David Crochemore RENATER France
Tiziana Ferrari INFN Bologna Italy
Joao Nuno Ferreira FCCN Portugal
Christoph Graf DANTE -
Aram Hayzapetyan YerPHI Armenia
Avgust Jauk ARNES Slovenia
Olav Kvittem (Chair) Uninett Norway
Simon Leinen SWITCH Switzerland
Ladislav Lhotka CESNET/USB Czech Republic
Olivier Martin CERN Switzerland
Kevin Meynell (Sec) TERENA -
Antonio Pinizzotto IAT-CNR Italy
Victor Reijs SURFnet The Netherlands
Esther Robles RedIRIS Spain
Vladimir Smotlacha CESNET Czech Republic
Pavel Smrha CESNET/UWB Czech Republic
Robert Stoy RUS/DFN Germany
Daniel Sturm CESNET Czech Republic
Andrei Sukhov SamaraNet Russia
Marc Thoelen Univ. Limburg Belgium
Celestino Tomas RedIRIS Spain
Jean-Marc Uze RENATER France
Apologies were received from:
Mauro Campanella INFN Milano Italy
Guenther Schmittner JKU/ACOnet Austria
1. APPROVAL OF MINUTES
The minutes of the final TF-TEN meeting held on the 2nd and 3rd of
July 1998 were approved.
2. STATUS OF QUANTUM & TEN-155
Christoph clarified the relationship between QUANTUM and TEN-155.
QUANTUM was the name of the project submitted to the European
Commission by a consortium of NRNs, whilst TEN-155 was the name of
the network. QUANTUM would also consist of a test programme similar
to TF-TEN.
A map of NRN access capacities to TEN-155 was displayed. These
ranged from 155 through 34 to 10 Mbps. Most lines were STM-1s, with
the remainder being E3s, and every NRN with the exception of Belnet
and NORDUnet would be using ATM as an access protocol. The backbone
would also be running ATM to provide both an IP and managed
bandwidth service. Initially only PVCs would be used, but it was
hoped that signalling could eventually be used.
The QUANTUM Test Programme (QTP) was still being determined. Each
NRN could decide whether to participate although it would be more
difficult for those NRNs that were not using ATM. Around 10% of the
TEN-155 bandwidth would be reserved for research purposes and it
would be possible to establish an overlay network similar to that
used for TF-TEN. Projects within the EU Telematics Programme could
also apply for bandwidth, although they would be charged for this.
The charging mechanism however, still needed to be worked out.
Initial experiments over TEN-155 were scheduled to commence on 1
December and would involved tests by mathematicians. The next phase
was reserved for QTP activities, and in the final phase, other
projects would come on to the network.
Olav asked what switches would be used in the TEN-155 backbone.
Christoph replied that Unisource planned to use Lucent switches.
Unfortunately, these had failed two acceptance test, and if they
failed a third later in the week, they would be rejected. It was
known that Cisco LS-1010s could fulfil the requirements, but these
were too expensive. In any case, it was the problem of Unisource,
not QUANTUM.
Olav also asked when TEN-155 was scheduled to start. Christoph said
deployment would commence in November with the production service
planned to start in December. Having said this, a contract had not
yet been signed with the European Commission (although this was
expected in the next couple of months). He added the TEN-34 contract
had been extended from June until December.
3. NATIONAL ACTIVITIES
Olav asked each NRN to provide an update on their current activities
and future plans.
3.1 Czech Republic
Ladislav reported little had changed since the last TF-TEN meeting.
They were still experiencing funding difficulties, although they
interested in taking part in the QTP. It was particularly important
to investigate QoS issues because they were currently running LANE
on the CESNET backbone.
3.2 Slovenia
Avgust reported their activities were hampered by the national PTT
that still had monopoly on telecommunications. They were quite slow
at providing high-speed links, and ATM was only available between 17
major towns. Even then, this was only used for leased line
emulation. Nevertheless, they were interested in taking part in the
QTP as they had suitable infrastructure in Ljubljana.
He added they were currently preparing the ground to use the
infrastructure of utility companies as soon as the monopoly of the
PTT was abolished on 1 January 2001.
3.3 Portugal
Joao reported they had negotiated free access to a JAMES-type
research network run by Portugal Telecom. This network consisted of
seven nodes and supported a virtual VP network, and a virtual
signalled network. This project was due to start in January 1999.
Unfortunately, Portugal would initially only have access to TEN-155
via satellite at 10 Mbps. This meant it would be impossible to
participate in the early QUANTUM tests. Hopefully they would obtain
an ATM connection in March 1999 that would allow them to join in the
experiments.
3.4 The Netherlands
Victor reported a few more ATM experiments would be conducted before
most of the ATM network was dismantled. These included IP over lossy
ATM, VBR overbooking and IP configuration on ATM. The backbone
network was being migrated to IP over STM-1, and would eventually be
upgraded to STM-4C.
CAR had been introduced on connections to the US in order to provide
certain institutions with reserved bandwidth in multiples of 128
Kbps. No performance issues had arisen as yet, but only basic tests
had been conducted. An MoU had also been signed with the Internet2
consortium and a Call for Tender had been issued for a 155 Mbps
connection to the STARTAP.
The Dutch government was currently discussing the SURFnet 5 proposal
for the GigaPort Project. This was planned to start in 1999 and
would run for four years. There were two aspects to this: GigaWorks
was an applications development programme in conjunction with
industry, whilst GigaNet aimed to establish a 2.4 Gbps research
network interconnecting SURFnet and telematics institutes. The
activity plan for this would be written in English.
Victor was asked whether he could circulate the URL of the activity
plan on the mailing list.
ACTION 981004-1: Victor Reijs
3.5 Italy
Tiziana said they were currently undertaking acceptance tests of the
new backbone; Milan, Roma, Bologna and Naples were being connected
by 34 Mbps lines. 155 Mbps lines were still not available so sites
had to apply for multiple 34 Mbps if they needed additional
bandwidth. There were plans to extend this backbone in 1999.
Cisco Stratacom and LS-1010 switches were being used in the backbone
to offer ATM with CBR VPs. A managed IP service was also being
offered, but there were no plans to use signalling. Government and
private research institutes would be allowed to conduct testing over
this network.
INFN was also taking part in the 6Bone and they were investigating
video-on-demand and QoS issues (RSVP, ATM and MPLS). Other research
projects included MONARC (distributed databases) and CONDOR
(distributed job processing between workstation farms). Finally,
they hoped to obtain funding from the EU Fifth Framework Programme
for an optical switching project. This meant they would be looking
for partners in industry and other research institutes.
3.6 CERN
Olivier reported CERN was looking to connect to Abilene in the US
and might be interested in participating in the QTP. It was mainly a
question of bandwidth and human resources.
They were also investigating video-on-demand and reflector
technology as they believed this worked better than multicasting.
Other developments included VRVCS which provided a Web interface to
the Mbone tools (unicast mode).
3.7 Belgium
Zlatica said she was not representing Belnet, but could report with
respect to VUB/ULB. Belnet did not have a test network, nor would it
have an ATM connection to TEN-155. This meant it would be difficult
for VUB/ULB to participate in the QTP, although they were looking at
alternative service providers. Nevertheless, they could conduct some
high-speed tests on-site as they were installing a 1.25 Gbps network
using the Ascom ATMLightRing 1250 System and Cisco LS1010 switches.
ATMLightRing is a ring-based ATM backbone network that uses optical
cell-based transmission equipment and permits nodes to be up to 60
km apart.
3.8 Switzerland
Simon reported the SWITCH NG project had been approved. The backbone
topology had been finalised and would be provided by diAx which
was a subsidiary of the power distribution companies. diAx would
provide an ATM service running at 155 Mbps, and SWITCH would operate
the IP service. Every institution would have access to both the
native ATM and managed IP services.
They were also thinking about the evolution of the network over the
next two years. It was hoped the network would be upgraded to 622
Mbps and would implement partial mesh UBR. In reality, it might be
advisable to use a full mesh as it was unclear how well some
switches handled congestion. At the same time, they would be testing
IPv6 and were looking for peering agreements with other IPv6
networks.
Unfortunately, there was still little interest in native ATM
applications except for the Telepoly project. This was very high
quality videoconferencing equipment based on the K-NET CellStack A/V
codecs that was being developed by the Swiss Polytechnics
(distributed institutions). It currently used 60-70 Mbps and they
hoped to offer this to other institutions.
3.9 Germany
Robert reported that DFN had procured a 2.4 Gbps test network
between Munich and Erlangen which had recently been extended to
Berlin. They were also hoping to bring Stuttgart onto this.
RUS had a local testbed for the EDISON project which investigated
QoS, IPv6 and IP multicasting. A MAN running at 622 Mbps was also
being established in the Stuttgart area that would utilise WDM
equipment from Bosch.
3.10 Spain
Celes reported they were currently using the ATM VP service of
Telefonica to interconnect seventeen regions in Spain. In two or
three months they would be moving to E3 connections, but still using
ATM. Unfortunately, the US connections were still overloaded.
They were currently investigating ISABEL (a high-bandwidth
videoconferencing application), IPv6 and RSVP, but would be unable
to participate in the QTP until April 1999.
3.11 France
Jean-Marc reported they were working on three main projects. They
had just issued a Call for Tender for a 155 Mbps connection to the
US, including 45 Mbps to STAPTAP. This would hopefully become
available in December.
RENATER II was also planned for mid-1999 and they had issued a Call
for Tender in June. This would interconnect twenty-five regions in a
proposed star topology initially at 155 Mbps, and later at 622 Mbps.
Both native ATM and IP would be offered, with multiple VPNs being
run over the ATM, and differentiated services being provided over
IP. Native multicasting and IPv6 would be progressively introduced,
and the backbone may use MPLS if production equipment became
available in time.
The SAFIR project interconnected nine research centres with 10
Mbps VPs for testing purposes. This was primarily being used to
obtain experience of running multiple VPNs over ATM, but was also
used to test PNNI, MPOA, NHRP and Telephony over IP. The contract
for this had recently been extended to add more sites and increase
the amount of bandwidth available. They were also trying to obtain
SDH or dark fibre to Eastern France in order to test WDM.
3.12 Russia
Andrei reported the financial crisis in Russia meant it was
difficult to find funding at the present time. Nevertheless,
SamaraNet had managed to obtain grants from NATO and other agencies
to enable them to build a 155 Mbps ATM network for academic and
scientific purposes. They were currently looking to upgrade the
connections with external networks.
3.13 Norway
Olav reported they had just put a new ATM backbone into service.
This used PVCs and mainly ran at speeds between 50 and 123 Mbps,
with remote sites getting 15-30 Mbps. Unfortunately, it would be
difficult to connect to the QUANTUM test network as they obtained
their international connectivity through NORDUnet (who would not be
using ATM). Alternatives were being investigated, but it really
depended on the cost of an access link to Stockholm.
4. DECISION ON NEW TASK FORCE(S)
Christoph said the meeting needed to recommend how the QTP should be
conducted, and whether a new task force should replace the TF-TEN.
Participants needed to consider that the QTP produced deliverables
for the European Commission and had to remain manageable and
relevant to TEN-155. On the other hand, some participants might wish
to undertake activities that were not defined in the QTP.
The intention was that activities in the QTP would be undertaken by
volunteers and would not be restricted to QUANTUM members. The
QUANTUM consortium however, had contractual obligations which made
it important that the Project Manager came from DANTE (who managed
the QUANTUM Project).
The proposed QTP activities were then displayed for discussion.
The meeting generally agreed with the activity list, but asked
whether changes could be made. Christoph said this was only a draft
document although the QUANTUM Policy Committee (QPC) needed to
approve the final programme.
Olav said he disliked the idea of an external committee controlling
the activities of a voluntary group, and felt there must be some
freedom to decide the programme. Christoph understood these
concerns, but said the QUANTUM consortium had to retain some control
to ensure their contract with the European Commission was fulfilled.
Olav asked about the membership of the QPC. Christoph was unsure
exactly who sat on the committee, but it was essentially comprised
of representatives from each member of the QUANTUM consortium. He
added that Howard Davies would be the manager of the QTP.
Victor and others said they had little interest in just working on
the QTP. This meant a structure had to be defined that would
accommodate other activities.
Kevin therefore outlined three possible formats. The first option
would be to form totally separate groups for the QTP and non-QTP
activities. The second option would be similar to the first, but in
recognition of the fact that both groups would have a similar
membership, they should hold joint meetings. The final option was to
form a joint TERENA/QUANTUM group similar to TF-TEN to work on all
activities.
The first option was unanimously rejected on the grounds that both
groups would have the same participants and would duplicate effort.
The second option was considered, but again rejected on the grounds
that effort might be duplicated. As a result, the third option was
accepted as the way forward, provided certain conditions could be
agreed.
It was recommended that a new task force should formed under the
auspices of TERENA in order to ensure autonomy from QUANTUM.
Nevertheless, QTP activities would be conducted as a major activity
if assurances were received that all the deliverables would be
placed in the public domain.
It was accepted that DANTE must have management control over the
activities of the QTP and, in recognition of this, they were offered
the Chair of the new task force with the proviso that sufficient
time would be spent on non-QTP activities. Christoph thought this
would be acceptable to DANTE so long as the focus of the task force
did not move too far away from the QTP. He then offered to take on
this role and the meeting was very pleased to accept his
candidature.
TERENA was asked to provide the secretarial support for the task
force. This would include the Meeting Secretary and the mailing
list. The WWW pages however, would be the responsibility of DANTE as
they already had to produce WWW pages for the QUANTUM Project.
The meeting asked for assurances that the new task force would be
party to all relevant technical information in the QUANTUM Project.
During TEN-34, information had sometimes been withheld from the
TF-TEN on confidentiality grounds and participants were told to
contact their managers if they needed access to this. In practice,
managers were always forthcoming with such information and it simply
wasted time. As members of the new task force would almost certainly
be representing NRNs anyway, this somewhat bureaucratic obstacle
should be removed.
The QTP activity list was modified slightly for clarification
purposes and potential participants were agreed as follows:
* RSVP - Simon Leinen (Leader), Tiziana Ferrari, Olav Kvittem,
Victor Reijs, Celestino Tomas.
* Multicasting (IP and ATM) - Robert Stoy (Leader), Tiziana
Ferrari, Ladislav Lhotka, Victor Reijs, Celestino Tomas.
* Differentiated Services - Tiziana Ferrari (Leader), Avgust Jauk,
Simon Leinen, Victor Reijs, Robert Stoy, Celestino Tomas,
Jean-Marc Uze.
* RSVP to ATM SVC Mapping - Tiziana Ferrari (Leader), Simon Leinen,
Celestino Tomas.
* IP Version 6 - Telebit (Leader), Tiziana Ferrari, Olav Kvittem,
Simon Leinen, Victor Reijs, Robert Stoy, Celestino Tomas.
* 155 Mbps ATM Performance - DANTE (Leader).
* ATM Signalling - Guenther Schmittner (Leader), Robert Stoy,
Jean-Marc Uze.
* Policy Control (IP and ATM) - Victor Reijs (Leader), Zlatica
Cekro, Tiziana Ferrari, Simon Leinen, Celestino Tomas, Jean-Marc
Uze.
* Route Monitoring - Simon Leinen (Leader), Jean-Marc Uze.
* Flow-based Monitoring Analysis - Simon Leinen (Leader), Victor
Reijs, Jean-Marc Uze).
* QoS Monitoring - Tiziana Ferrari (Leader), Olav Kvittem, Victor
Reijs, Robert Stoy, Celestino Tomas, Jean-Marc Uze.
The group also proposed additional activities that were not included
in the QTP:
* MPLS - Jean-Marc Uze (Leader), Tiziana Ferrari, Simon Leinen,
Ladislav Lhotka, Celestino Tomas.
* VPNs - Victor Reijs (Leader), Robert Stoy, Celestino Tomas,
Jean-Marc Uze.
* WDM - Victor Reijs (Leader), Tiziana Ferrari, Jean-Marc Uze.
* STM-1 vs STM-4(C) - Victor Reijs (Leader), Simon Leinen.
Christoph was asked to produce a policy document summarising these
technical recommendations that could be presented to the QPC. He
agreed to do this.
ACTION 981004-2: Christoph Graf
It was also agreed that each activity leader would produce an
experiment proposal for further discussion.
ACTION 981004-3: All activity leaders
Victor raised the question of how the activities would be
written-up. Obviously the results of the QTP activities would form a
deliverable to the European Commission, but this did not apply to
other activities. It was agreed these results would be published as
a TERENA Technical Report.
Olav asked whether the new task force should retain the 'TF-TEN'
title, or whether a new title should be used. It was agreed this
issue should be discussed on the mailing list if and when the task
force was ratified by QUANTUM and TERENA.
Finally, Jean-Marc asked why Telebit A/S were a member of the
QUANTUM consortium. Christoph replied that EU projects were supposed
to include a partner from industry.
5. DATE OF NEXT MEETING
The first meeting of the new task force will be held on the 5th and
6th of November 1998 at the DANTE Offices in Cambridge, UK.
6. ANY OTHER BUSINESS
Kevin mentioned that TERENA currently had a vacancy for a Project
Development Officer. Further information was available from:
http://www.terena.nl/info/jobs/
OPEN ACTIONS
981004-1 Victor Reijs to send URL of GigaPort Activity Plan to the
mailing list.
981004-2 Christoph Graf to produce policy document summarising the
technical recommendations of the WG-LLT.
981004-3 All activity leaders to produce experiment proposals.
Contact: nep@dante.org.uk