Editor's note: These minutes have not been edited. Roger Fajman - Conducting Bill Manning - Minutes Agenda * Responses to solicitation RFC - Will Leland Renumbering Case Study - Elliot Lear (SGI) * Vendor Solicitation - Will Leland * Router Renumbering ( HWBerkowitz - by phone or Paul Ferguson) * Network renumbering overview (Paul Ferguson) * How to renumber a small site * IP addresses in applications (Phil Nesser) * Graceful renumbering of networks w/ DHCP (Lowell Gilbert) Several Case studies have been submitted. All were lacking in details. It is tough to get documentation since some consider it confidential and others don't create the documentation before or during transition and don't budget the time to do so afterwards. These were also the reasons why it is difficult to get documentation or details on the tools people use to assist in renumbering. Most consider these scripts to be "throw-a- ways" or one-offs. There was a call from some attendees for more experiences in the use of NAT/ALG technologies as renumbering techniques. There was also a question raised on the effects of firewalls on renumbering efforts. Will has collected many studies in the following URL. It will be replicated on the PIER web page. ftp://ftp.bellcore.com/pub/wel/PIER/input Elliot Lear pointed out a social tool for reengineering, bribary and "gimies". SGI is not yet complete with its renumbering effort. The main driver was to clean up routing. A foraml analysis is being done for the LISA-10 confernece and will be made available to the IETF. In the appendix, documentation of tools will be provided. There is extensive use of DHCP and utilizing vendor support teams for planning. SGI found that it took about 50% of the time in prior planning. There was the need to clearly understand routing protocols and route distribution as protocols were changed to support variable length subnets. SGI is not planning to renumber again anytime soon. subnets were allocated based on corporate growth plans. It is expected that there was an increase in the percentage of utilized numbers. SGI developers became involved and the SGI OS will support variable lenght subnet masks and RIPv2 in the kernel sometime in 1996. At least this is expected. SGI found that renumbering may have two foci; local aggregation or wan aggregation. One is pushing a bit mask about, the other is prefix changes. There are interactions between frame relay subinterfaces and number of OSPF links. Point to point frame relay subinterfacees used IP unnumbered stratagies in transition only. Transition time per site was between two and eight hours per site. The time was a direct ratio of variance from corporate standards. Will Leland presented an update on the merged MCI/Droms draft. It was agreed to send it in as a draft, wait minimum period and then ask for FYI RFC status. Paul Furgeson reviewed his network renumbering draft. Its a real draft, ready to go. It was suggested that it be revised to incorporate some of the statements from Howards draft and add RIPE 136 to the summary, then be reposted to the list for comment. We would like to send this to the IESG for last call by mid-July. If there is time, and information, Paul might wish to add a section which clarifies the use of route summarization. A brief discussion of Howards document. The vendor appendixes will be removed due to lack of input. The overview section will be moved to Pauls document. Howard will add a summary section and then procced with WG last call and then go for FYI. A document on "How to renumber a small site" still needs to be written. Bill Manning agreed to do so. Phil Nesser had zero input on imbedded IP addresses. It was covered in the session that there are existant licensing issues and other code points but we don't really know specifics. Some general areas are: Firewalls,ACLs, Eeprom OS'en, SW License servers, TIS toolkit, DNS cache, Wais cache, HP Openview & any/all SNMP mgmt stations, NFS configurations, NATs etc. The PIER list should work with Phil so he can have updates by 01aug96 and have a draft of some text shortly there after. Lowell Gilbert: DHCP. as a gracefull renumbering tool. Its current status is that it is being progressesed in the DHCP. Rocky Flats (CO) has done an implementation. It does need a host and server implementation. We will progress this. It is noted that this does change host behaviour. It is expected that we will issue a last call to the list and then ask the IESG for experimental status.