DRAFT minutes of FInal RAMS meeting of 4 Dec 1996 (draft of 19/9/96, by Jon Crowcroft) Meeting Held at UCL, London, 11-4 Agenda - agreed as follows: 0. Minutes of previous meeting(s) and agree agenda 1. Final Deliverables status (see attached for list) 2. Final Report Review 3. Claims status 12.30 Lunch 4. Project Open Day Plans 5. Other Explotation Plans 6. AOB Attendees --------- Mike Falla, mike.falla@bcs.org.uk _ Monitoring Officer Steve Osselton, Prism, steve@prismtech.co.uk Tim Snape, Abbotsbury/WDI, tim@wdi.co.uk Jon Crowcroft, UCL, jon@cs.ucl.ac.uk James Cowan, J.Cowan@cs.ucl.ac.uk David Romano-Critcley, UCL Kieran Cooney, Texaco, udmkxc@texaco.com Huon Butterworth, Tadlon Apologies from Robin Crosher, DTI, robin.crosher@dtied.dti.gov.uk - HPIP Program Manager Deborah Miller, EPSRC 0. Minutes from last Meeting and Actions Arising ----------------------------------------------- Minutes approved. Action: Jon to send out corrected version. 1. Deliverable Status --------------------- All UCL deliverable documents will be avaialble through our Web server (http://www.cs.ucl.ac.uk/staffjon/rams/rams.html), and as Research Notes from the departmental librarian. We shall also be making the managed protocol stacks for Nt and Linux available in the public domain (and possible through the main official Linux release mechanism) together with a suitably instructive, but not full functional demonstrator to show how to use the API. We then went through the full list deliverables to establish their status, and prognosis. Bracketed [] deliverables will not appear in the list in the final report, but are included here for information. in the final report, proposed and actual delivery date will be listed, along with effort planned and expended. UCL --- Deliverable UCL-1: report due end of March 96 Unix protocol stack management stack specification A specification of what variables in a Unix protocol stack. done Deliverable UCL-2: report due end of June 96 Winsock protocol stack management specification Design of the managed protocol stack for a Winsock tcpip stack. This will describe how the stack is monitored and managed and how connection scheduling will be implemented. done Packet Analysis Deliverable: report by Tadlon With Tadlon with the traffic analysis of Texaco's network. done Deliverable UCL-3: report due end of July 96 Network performance analysis Analysis the delay variations within Texaco's LAN and WAN. done [Deliverable UCL-4: software due end of July 96] [Winsock monitoring software] [A spy program that can track the performance of connections within ] a TCP-IP winsock stack. This work will be a deliverable for the BT project.] [Note that this is a BT funded deliverable, not a RAMS funded deliverable.] Deliverable UCL-5: software due end of December 96 Winsock stack management Connection scheduling to the Winsock management software. done - but note key problem here: The Tadlon final deliverable (see below) requires us to run a SAP server and compare and contrast the monitored/management of the stack with the packet statistics from Texaco's network. Unfortunately, UCL cannot solve the problem of running the winsock managed stack except with Microsoft Windows Nt 4 (and Linux), but Texaco do not plan deployment of SAP server side systems on Nt 4 until Q2 '97 (at earliest - contingent on SAP delivery!). We could do client side Nt 4 tests, but this would not reveal a lot about the system. An alternative deliverable for Tadon is proposed below using the Nt 4 or linux setup with a Web server ratehr than a SAP server, t WDI instead of Texaco. We will still try to carry out the client side work at Texaco, and if there are resources (UCL and Tadlon are willing, and Texaco are happy with this plan), we can try the origianl experiment later in '97 as part of our exploitation work. Deliverable UCL-6: software due end of December 96 Network management software A network management application in the JAVA programming language that has a GUI that displays connection information and allows network managers to control the stack of a PC. in hand - Dave Romano-Critchley gave a presentation about this in some detail - this will be part of UCLs talk and presentation at the Project Open Day. [Deliverable UCL-7: report due end of November 96] [SNMP MIB] [We may design an SNMP Management Information Base that would form the ] [basis of Internet standard for managing TCPIP connections. James Cowan ] [would write this MIB (depending on agreement with BT).] [Note that this is a BT funded deliverable, not a RAMS funded deliverable.] done Prism ----- Deliverable Prism-1: Design document due end March 96. Resource/Environment model design done Deliverable Prism-2: Design document due end March 96. Resource manager design done The following bracketed () deliverables were staging posts along the way to developing Prism-11. As such they were internally delivered, and will all be considered complete by the Project Manager when the full demonstration is done at the Project Open Day. (Deliverable Prism-3: Software due mid May 96.) (Resource model implementation ) (Deliverable Prism-4: Software due mid June 96.) (Resource manager implementation ) (Deliverable Prism-5: Software due mid August 96.) (Integration with OpenBase runtime ) (Deliverable Prism-6: Software due mid October 96.) (Extension of OpenBase CDL ) (Deliverable Prism-7: Software due end July 96.) (Extensions to OpenBase meta model ) (Deliverable Prism-8: Software due end June 96.) (Implementation of static resource loader ) (Deliverable Prism-9: Software due mid September 96.) (Implementation of dynamic resource monitor ) (Deliverable Prism-10: Software due end June 96.) (Resource monitor ) Deliverable Prism-11: Demonstration at end of project. Demonstration system To appear at Project Open Day. Deliverable Prism-12: Extended OpenBase documentation due at end of project. Documentation Done, but not delivered. UCL and the Monitoring Officer asked for copies of these to be sent. Steve O. said he would do this right away. Action: Steve Ossleton - deliver documents to UCL and M.O. Tadlon ------ Deliverable Tadlon-1: Baseline data and reports. First Analysis Phase. done Deliverable Tadlon-2: Baseline data and reports. First Analysis Phase. done Deliverable Tadlon-3: Detailed report of the traffic on Texacos network comparing results both phases and giving our conclusions. Produce report detailing conclusion. This will now be done in week of 16 Dec, using Linux or Nt 4 web servers at WDI/Abbotsbury - initially, we will deploy the system at UCL, and once running transfer this to WDI, and then write the report in time for the new end of project date. Action: UCL, Tadlon, Abbotsbury - arrange alternate Tadlon work Abbotsbury ---------- Deliverable Abbotsbury-1: Operation web site, logs of activity. Establish Web site. done Deliverable Abbotsbury-2: Draft quality manual, the deliverable will be created in hypertext format. The manual submitted to BSI will be a printed manual. Produce a quality manual Done, but Delivered Action: Tim Snape - send copy draft QM immediately to M.O. and UCL Deliverable Abbotsbury-3: Report on Fault Expert Still to be done. Deliverable Abbotsbury-4: Final Quality report Very nearly there. Deliverable Abbotsbury-5: Justification of Quality Procedures - due dec 11 in progress in addition, abbotsbury have sub-divided the work on the QM into internal and external procedures, and will be able to make these reports avaialble in that structure. 2. Final Report --------------- The M.O. had provided a very useful outline document which we went through to see what actions were needed to provide a full final report in good time. This will be commercial in cofndence and goes only to the M.O., the DTI program manager and staff, and the EPSRC program manager. Note that UCL also have to produce a report to the EPSRC. i) Summary. The Resource Allocation Management Systems (RAMS) Project was a 1 year program of research and development which sought to produce a set of software systems that could be used together to ensure that critical applications have the network resources they need to operate efficiently. The target networking environment was specifically the Internet. The resulting system has two main components: manipulation of the protocol stack software on individual machines, controlled to a resolution that affects the traffic injected into the network by different applications. The other approach will involve controlling the application directly in order to change the volume of network traffic it generates. ii) Collaborators: A list of the companies and universities/polytechnics involved and their expenditure on the project. The partners in the project were: University College London, Computer Science Department - Lead, EPSRC funded. 1.5 full time research fellows Prism Technologies Ltd. 1 full time engineer Abbotsbury Software Ltd 1 full time engineer, with a sub contract to Tadlon and Texaco A user company, contributing expertise, and a real world live test network. iii) Aims and Objectives: A brief statement of what the collaborators were trying to achieve in the project. Jon will do... iv) Technical Achievements: What were the original main technical objectives (milestones, deliverables) and to what extent were they achieved? Were any initially unforeseen problems overcome during the project. Are there any technical problems not solved which might demand further attention? Action ALL - provide 1/ page over hext 10 days on their technical achievements for Jon to incorporate Web page papers presentations s/w collabs.... v) Effectiveness of the Collaboration: Was the collaboration effective (illustrate with a few examples)? Would you do the project this way again and would you collaborate in future projects? The project benefited from a careful partitioning of the work. The UCL work was central hub, with spokes out to Texaco and Tadlon, to carry out network monitoring, to Abbotsbury for Internet Service Provision and Quality Monitoring, and to Prism for load monitoring and balancing. This meant that there were few cross dependancies, if any. We have used this structure before, and would use it again. Mike Falla noted that it is important not to imply that there might have been seperate projects - Jon will stress that the feedback from SAP and Prism and WDI requirements were all useful for the UCL Stack work. vi) Constraints: Were there any barriers to good collaboration? How could these be overcome in the future? In this work, the only threats to our work have been external - typically from lack of adequate support from Microsoft. I'll add material here about problems of depending on a large system deployment (SAP on Nt ) in a 1 year project.... ote we were NOT hindered by IPR, non-disclosure, or other barriers which was very nice - we might have had problems publishing detailed results about SAP externally, but this won't invalidate the project one iota, since the information will still be of immense value to Texaco and Tadlon. vii) Other Outputs: Were there other outputs from the project eg publications, patents or other technology transfer? Provide a brief summary (details in an annex). Papers....Internet Drafts....Software products.... Action all: any other things output.....for final report viii.Exploitation: When will the results be exploited, eg in the manufacture of goods or to improve processes or measurement procedures? If exploitation is not anticipated now, explain why not. What further work is needed before commercial exploitation can begin? Are there plans to carry out this work? Would another company not involved in the project be able to exploit the technology, by licensing if necessary? Abbotsbury + ISPA..... Prism + Products UCL + Texaco case study... Action all: report on exploitation plans to jon...... ix. Impact of project: Can you place the results of the project in a world wide context? Where does the UK stand in relation to overseas competitors and how has this project helped in a significant way? Hard to estimate UK now has 870,000 hosts on the Internet. With 300 providers, we have a lot of good expertise. In Orbs and distributed Systems technology, there are a small number of outfits...its not clewar how well placed we are (Iona in ireland, HP labs, Tibco/Technkron, etc etc) Jon will collate this part. ISPs quality.....etc etc x. Other comments: Unexpected bonuses from the project, etc. No unexpected positive bonuses. 3. Claims status ---------------- The only missing 3rd quarter report was Abbotsbury due to a postal and fax mixup - this has now been rectified. Jon stated firmly that he would be pursuing final claims very avidly. 4. Project Open Day Plans ------------------------- i) Outline Program 10 Coffee 10.30 Open 10.30-10.45 Introduction to UCL/RAMS - Jon Crowcroft 10.45-11.00 Mention ISI, HPIP, partners, and outline day/purpose 11.00-11.30 UCL "Managed Protocol Stacks" 11.30-12.00 Abbotsbury High Quality Internet Service Provision 12.00-12.30 Prism - Monitored and Managed Distributed Systems 12.30-1.30 buffet lunch 1.30-3.00 Structured Demonstration Talks in room 229 in parallel with Open Demonstrations in 214 3-4 open demonstrations continue in 214 with cofee/tea i) Invitation Will send a draft of this - action jon- generate Open Day invitation and send to partners iii) Invitees UCL will invite Mike Falla, project M.O. Robin Crosher + any colleagues he wishes to bring Peter Burton Deborah Miller from EPSRC UKERNA representatives BT labs, Microsoft UK, FTP Software HP Lbs Reuters UK Other HPIP program project reps Prism to report who they will invite fro mcustomer base Abbotsbury Will invite ISPA cttee people and customers 5. Other Explotation Plans -------------------------- UCL reported on putting managed stack in Public Domain. Report on texaco traffic will also be made avaialble as case study - we will liase with Abbotsbury about the QM. 6. AOB ------ Deliverables i) Project End Date The DTI offered us an esxtension, which Prism, Tadlon and Abbotsbury have decided would be useful - we await formal confirmation, but informally, it has been edxtended til January 21st. The EPSRC have confirmed UCL's extension til end of January 97. Neitherextension incurs extra costs to DTI/EPSRC, nor extends the "no obligation to pay date" at all. ii) Tim Snape Reported that on the 5th Dec, they (WDI/Abbotsbury) would learn about their "Rural Challenge" Bid iii) Jon Crowcroft mentioned the Digital Broadcast Link program between EPSRC and DTI, and that UCL are a member of the Digital Broadcast Virtual Center of Excellence - UCL would welcome proposals to form a consortium to bid for Link money. 7. Actions Arising ------------------ Action: Jon to post last meeting's corrected minutes done. Action Jon: to collect and collate full setr of deliverables Action ALL: to send Jon latest deliverables asap after he requests them Action: Steve Ossleton - deliver Prism documents to UCL and M.O. Action: UCL, Tadlon, Abbotsbury - arrange alternate Tadlon work Action: Tim Snape - send copy draft QM immediately to M.O. and UCL Action ALL - provide 1/ page over hext 10 days on their technical achievements for Jon to incorporate Action all: any other things output.....for final report Action all: report on exploitation plans to jon...... Action UCL: To produce final report final draft by dec 16 action jon- generate Open Day invitation and send to partners