don@SRI-LEWIS.ARPA (Donald Holman) (12/10/87)
To those that listen with appologies to those that don't: I left the TCP/IP conference (Monterey East) with a few items that appeared to be without solution in the near future. I query the mailing list in an effort to solicit for ideas from the collective body on each of these areas. One or all of these will be of interest to my efforts in the near future, thus why wait for the rabid bite of a problem without solution. Probably be inundated with flaming broadcast storms over these questions, but here goes. PLEASE; IF YOUR RESPONSE IS NOT ENTIRELY APPLICABLE TO THE TCP/IP MAILING LIST RESPOND DIRECTLY TO ME. Glossary of terms and acronyms: --------------------- Standard - an accepted degree or level of excellence. TIA - That Is All. 1) SLIP LINK ESTABLISHMENT, there is no standard link establishment procedure/protocol for SLIP which presents problems when attempting to utilize SLIP in a dialup or various other modes. There are additional problems which are related to address mapping for the user of a SLIP port, which incidently may cause problems with security. Is anyone working on a solution to this, or is there an existing solution may have the pleasing aroma of a standard? 2) SECURITY, various vendors of IP routers touted the ability to discriminate for/against packets based on the security field found in the IP portion of a datagram header. The present problem with this is that there is no provision in any of the ARPA/DDN upper layer protocols for setting this security bit (or is there?). How does one determine to set the bit since SMTP, FTP etc. does not provide the mechanics for passing this security information to the network layer? 3) ROUTING METRICS. Most gateway routing decisions are determined by hop count or minimum path. This is not always the best approach since the minimum path may not be the best path. Example: who cares if the packet traverses 5 hops to get to the destination if the 5 hop path is more economical/better than the 2 hop path. Is there a published method that is better than min_hop? 4) LEASED LINES FOR DDN. The DDN may not be able to afford all of the dedicated leased lines which are to be in service over the next several years. Is there an alternative to leased lines? Perhaps something that utilizes dial-up in conjunction with a machine-machine (read uucp-uucp) transfer of queued information. This could reduce the need for high-cost leased lines in selected cases. TIA, Don