BT/JISC University Research Initiative Management of Multi-Service Networks Management Meeting No. 3 Location: University College London (UCL) - 14 September 1995 Present:- BT UCL Cambridge Lancaster Ian Marshall (Chairman) Jon Crowcroft Ian Leslie David Hutchison Mick Mulvey (Secretary) James Cowan Simon Ritchie Lougborough Oxford Brookes IC David Parish Sheila Lloyd Lyons Morris Sloman JISC UKERNA Ron Rogerson John Dyer 1. Use of SuperJanet Infrastructure Jon Crowcroft gave a brief histroy of the SuperJanet network evolution. Of the six partners, Cambridge, ICL and UCL are connected to the ATM and SMDS networks; Oxford Brookes, LUT and Lancaster are only connected to the SMDS network. The ATM network had initially comprised a number of GDC DV2 ATM switches interconnected via 34Mbit/s PDH line systems. Initially the GDC DV2 had only single QOS output buffers, these have since been upgraded to dual QOS to eliminate 2Mbit/s circuit emulation problems due to cell loss when contending with highly bursty VBR traffic. More recently Fore systems ASX200 ATM switches had also been connected into the ATM network these were thought to have better capabilities for traffic partitioning. In Jon s view all of the upgrades and enhancements to date had come about from a need to optimise the service side of the SuperJanet network as opposed to facilitating network research. He also noted that all the work so far undertaken as part of the SuperJanet URI had been done in isolation to the Superjanet ATM network. Examples given were the work undertaken on the PNO ATM pilot and work performed by Lancaster on the their own ATM infrastructure. It was hoped that from here on the ATM infrastructure would be sufficiently stable to allow URI research to make use of the SuperJanet ATM infrastructure, particularly with regard to CBR/VBR traffic policy experimentation and possibly, later in year 3 of the URI, ABR experimentation. However, this would be dependent on the stability of the ATM infrastructure over the next 2 years. David Hutchinson s suggestion of a dedicated research network was not supported by Ian Leslie who thought that a long term strategy for the provision of virtual networks within the ATM infrastructure would be a much better target to aim for. This was agreed and Jon suggested a need for a technical policy of traffic quarantineing to ensure traffic contention between concurrent research and service traffic does not cause service disruption. As an example, Jon was keen to try out a multi-site VBR video conference demonstration but there appeared no easy way at present for organising the network configuration to allow for this. In response to David Hutchinson s additional suggestion that requirements be targeted at JISC directly, Ron Rogerson indicated that any such initiative would have to be in the form of a detailed proposal, and that the universities should bear in mind that the ATM network currently in place may bear little resemblance to what the network may look like post 1997. Jon Dyer provided further details of network stability and the likely migration strategy for the ATM network. The ASX 200 switches had been provided to 5 sites, UCL, IC, Rutherford, Manchester and Birmingham, to allow connection to the planned SDH line systems. The SDH line systems are currently interconnected via an AT&T ATM switch based on BT premises at Ealing. However, this switch had no formal maintenance contract and only operated on a wing and a prayer Additional upgrades to 155Mbit/s at other sites had been delayed. As an interim solution 3*34Mbit/s connectivity was being provided. It was not yet clear when the 155Mbit/s lines would become available and what ATM switches would be procured for the SDH lines once they are in place. Currently Ukerna were adjudicating switches from a short list of 4 companies with examples of these equipments currently being tested at UCL and Edinburgh. The results of this adjudicated should be known at the end of September 95. John Dyer indicated that from Ukerna s view point the ATM network does not yet offer a formal service but will do on the 1st of April 1996. It was agreed that to enable such research to take place on the ATM network some form of booking system would be needed. John Dyer again (see Action Point 4 from previous management meeting 28 March 1995) indicated the existence of a document concerned with the procedure for applying for bandwidth on the ATM network. The three universities connected to the ATM network, Cambridge, UCL and IC, agreed to comment on this document and construct a proposal for the allocation of bandwidth for the URI project research. It was noted that this proposal would be in two parts. An initial part highlighting the URI s immediate requirements would be available within 6 weeks. The second part concerning the long term strategy of virtual networks would have to follow some time after. AP-1/UCL/IC/Camb. John Dyer added that the ATM network is currently managed by the Ukerna technical group based at Edinburgh. However, this management responsibility is soon to be handed to Ukerna network services unit and may be relocated . 2. Deliverables From First year. Ian Marshall worked through a comparison of the deliverables described in the deliverables document (dated April 17th 1994) and the actual deliverables provided as an output for year 1. It was noted that any changes in deliverables had been as a result of redirection by and in agreement with BT. The following list of deliverables and actions were agreed:- 2.1 Imperial College Original Deliverables:- Report on Configuration Management for Multiservice networks and Demonstration of configuration management tools, Report on Policy Notation for specifying traffic management policies. Actual Deliverables Submitted:- Two papers on configuration management had been provided to BT and a toolset demonstrator had been developed. It was noted that the contents of the papers were based on previous intellectual property but that the contents of the demonstrator were wholly owned by BT. Morris Sloman agreed to forward the titles of the papers to John Crowcroft for inclusion in his summary report of year 1 activities. AP-2/IC/MS 2.2 UCL Original deliverables:- Policy Specification Notation, and Automated Traffic Management System Actual Deliverables:- Four reports have been submitted to BT and software has been implemented on the PNO pilot. All four reports are fully owned by BT. Jon would include the exact titles in the year 1 report. AP-3/UCL/JC 2.2 Lancaster Original Deliverables:- . Reports on QoS Specification and support mechanisms, Report and demonstration of QoS controlled Multimedia storage device, report on congestion and control mechanisms Actual Deliverables:- A total of four documents had been provided as deliverables - a review of QoS architecture, QoS management in end systems, methods and mechanisms for QoS multi-cast groups, QoS guarantees across heterogeneous networks. All reports except the second were wholly owned by BT. David Hutchinson agreed to submit to Jon Crowcroft an indication as the percentage of BT ownership of the second report. AP-4/Lanc/DH 2.3. OB Original deliverables:- QoS Specification and Support mechanisms document, multimedia storage specification document, algorithms for congestion control mechanisms. Actual deliverables:- A software simulation of an Available Bit Rate Network had been completed and was now ready for undertaking bandwidth allocation studies for the next years. Two papers had been published. Sheila Lloyd Lyons agreed to forward details of the title of these papers to Jon Crowcroft for inclusion in the year 1 summary report. AP-5/OB/SLL. Both the software model and Intellectual property contained in the two papers are wholly owned by BT 3.4 LUT Original deliverables: Mulit-rate Video and Audio service specification, Traffic monitoring and measuring toolset, early video demonstration experiment. Actual deliverables:- Four papers had been delivered to BT, an ATM over ATM discussion document, a detailed performance measurement architecture document, a paper presented at Coseners Multi-Service Networks conference, and further paper in support of a potential patent application. Two hardware measurement systems had been developed one an enhanced PING based measurement tool, and the other a dedicated unidirectional SMDS measurement system incorporating GPS timing. Much measurement data from the PING measurements and dedicated hardware system had also been transferred to BT. All of the intellectual property of the above is wholly owned by BT. Titles of the papers would be sent to Jon for incorporation in the year 1 report AP-6/LUT/DP 3.5 Cambridge Original deliverables:- Security Model and implementation for signalling, Implementations of end-to-end security mechanisms. Actual Deliverables Three papers had been delivered to BT, one outlining the assumptions forming the background to the work, a second comprised a survey of security methods, and finally a plan for prototype implementation. Titles of the papers would be sent to Jon for incorporation in the year 1 report AP- 6/LUT/DP Ron Rogerson indicated that JISC had no comments to make on last years deliverables. 4. Work Plan for Years 2 and 3 of the Project The following amendments to the years 2 and 3 plan of work were agreed:- Brief statements outlining the expertise of the Principals and Research Assistants involved in the project would be included. Statements on package linkages would be included in each of the workpackage descriptions and also in a separate linkage section which would pull all of the linkages together. IC s comments on insufficient funding within the URI to complete the work described would be reconsidered and only work appropriate the URI and wholly funded by it would be included in the Year 2/3 plans. UCL s deliverables would be made more clear, for example statements on when prototype source code would be available for analysis would be included. Lancaster s section would include statements on what intellectual property would be transferable to BT. LUT were commended for having the most detailed set of deliverables. All universities would check with their finance departments that they agreed with the figures quoted. AP7/All Universities. The initial part of the document would make clear that this is a continuation of the SuperJanet URI contract. Work package 5 would be moved in front of work package six. Oxford Brookes deliverables would be reformatted to align with the rest of the work packages. Year 3 costs would be brought separately and in a common format on all work packages. Cambridge s deliverables would be re-worded to specify the delivery of reports. 5. Any Other Business 5.1 Mail Lists - Ian Marshall agreed to set up a local BT mail exploder to enable interested BT personnel to receive URI material. Jon Crowcroft would be advised the e-mail address of this exploder for addition to the technical and management mail lists and a list of main BT personnel subscribing to the exploder. 5.2 Meetings- It was agreed that the frequency of the technical meetings would be increased the quarterly to better facilitate the interworking between work packages. 5.3 VC8000 N-ISDN video conferencing Simon Ritchie queried whether their would be any benefit to the project, specifically work package interworking , in using VC8000 N-ISDN video conferencing tools. However, the universities preferred to look into the use of the M-Bone network, which additionally provided muli-point conferencing, and also to investigate VBR video conferencing capabilities which promised better quality operation. The target would be to have the third technical meeting over the M-bone provided the universities could figure out the logistics of organising this. 6. Date of Next Meeting The next Technical Meeting will be held on Wednesday 13th December at Imperial College commencing 10:30am. In the evening it was agreed that there will be a URI dinner - to be organised by Imperial College but funded individually. The next Management meeting will take place on Tuesday 19th March at UCL. This will be a joint technical and management meeting. END OF DOCUMENT