CURRENT_MEETING_REPORT_ Reported by Andy Nicholson/CRAY Minutes of the Dynamic Creation of Network Links BOF (DCNL) Description: We held BOFs on this subject at the 20th and 21st IETF's, under the name ``Conditioning of By-Request Network Resources''. This is a continuation of that interchange. The change of name was suggested by Noel Chiappa. This meeting attracted much more interest than the previous meetings. Attendees generally suggested developing a charter for the possibility of starting an IETF Working Group. While working with circuit-switched T3 networks, developers at Cray Research Inc., determined that there would be advantages to defining a standard way to control certain classes of network resources through the internet. In the case of a circuit-switched T3 line, the line should be switched on only when there are active transport connections which can fully utilize the service. Due to the high cost of the resource, underutilization would be particularly undesirable. The developers believe that this capability might have other applications in the internet and that an effort should be made to define a standard protocol. Minutes: The meeting began with a presentation by Andy Nicholson regarding the work done at Cray Research with circuit-switched T3 networks. This was a review of the internet draft draft-nicholson-conditioning-00.txt. This was followed by a short discussion of the Link Control Protocol used by the Cray Research demonstration software. This protocol is mentioned in the existing draft and will be fully described in an upcoming internet draft. The attendees discussed different methods of supporting this service and how it might fit into the infrastructure of the internet. One possibility is that rather than transport providers deciding when to activate and deactivate links, intermediate routers in the network may decide to perform this function when there is a need. In this way the network could automatically adapt to changing network load and delay conditions. Some of the attendees were suggesting other possible uses of this capability and some discussion ensued. The most likely other use of this type of support for circuit-switched links would be for planned capacity management where an administrator might bring extra capacity 1 on-line in the network during peak load times. Other possible uses are for backups to existing primary links or for bypass links when there is sufficient traffic between distant endpoints to avoid multiple hops between those endpoints. Bill Jolitz suggested considering the management of the dynamic links. Rather than developing a new protocol (LCP) to create and manage links, SNMP could be used if a proper MIB was developed. He went on the suggest that this could be used as a methodology of defining the work to be done by making a first cut at a MIB for this facility. A Working Group could then iteratively improve the MIB while refining the functionality of dynamically created links. We also discussed the various issues presented in a slide, and the attendees suggested other issues requiring consideration. One issue is that when bringing extra bandwidth into the network to alleviate Congestion, the relief may only be temporary. As senders discover the extra bandwidth, they may increase their output to use it up. Another possibility is that the routing protocols may dynamically change the network topology to suit the changing demands, and this would add new complexity to routing and routing protocols. Ken Hayward was concerned whether this service would have a useful lifetime, considering that there does not seem to be, at present, an analog to switched T3 in the ATM world, and that ATM might address the issues presented here. There was general agreement that this was a good point, but that we could not predict the future. It was further noted that some networks might wish to have dynamic control of slower links, such as in the case of bringing backup links on-line when a primary link fails. The attendees generally agreed that this is an interesting topic of discussion and expressed a desire to see a concrete description of the problems to be addressed. I agreed to develop a charter which addresses these concerns. I also agreed to install a mailing list for discussion of this topic. If there is sufficient interest then I will present the charter to the IESG for Working Group status. At the meeting I described a paper published by CICNet in July, 1991, which mentioned their interest in circuit-switched T3 services. I promised to include instructions on how to get this paper. That information is presented here. A report titled ``High Performance Applications on CICNet: Impact on Design and Capacity'' is available from CICNet via anonymous FTP. {\bf host:} NIC.CIC.NET 2 {\bf directory:} /pub/reports {\bf file:} ds3-report.[ps or txt] ABSTRACT: This twenty-three page report summarizes available network technologies, reports on a survey of the needs of researchers and faculty at CIC institutions, and provides detailed studies of network requirements in four areas of contemporary, scientific research. The needs of these four areas of research are then summarized in terms of network requirements, and specific recommendations are presented by the Working Group to CICNet, Inc. The report was authored by the CICNet DS-3 Working Group, which was chaired by Mike Enyeart of Indiana University. [Final Note: A mailing list for this work has been set up, its address is dcnl-ietf@cray.com. Attendees Jordan Becker becker@nis.ans.net Richard Cherry rcherry@wc.novell.com Erik Fair fair@apple.com Craig Fox foxcj@network.com John Gong jgong@us.oracle.com Kenneth Hayward crm57d@bnr.ca Greg Hollingsworth gregh@mailer.jhuapl.edu William Jolitz william@okeeffe.cs.berkeley.edu Andy Nicholson droid@cray.com Timothy Salo tjs@msc.edu Stephen Shew sdshew@bnr.ca Martha Steenstrup msteenst@bbn.com Brad Steinka brad@python.eng.microcom.com Preston Wilson preston@i88.isc.com Robert Woodburn woody@cseic.saic.com Richard Woundy rwoundy@ibm.com 3