hedrick@TOPAZ.RUTGERS.EDU (Charles Hedrick) (08/09/87)
Sorry, I should have been more clear about the phase V DECnet addressing. First let me note that I am basing this on a paper given a few months ago at an IETF meeting. I have not seen full documentation for phase V. According to this paper the basic DECnet address will involve 2 bytes of area number and 6 bytes of system ID. "DEC uses Ethernet absolute host ids in this field to ease address administration. The only requirement, however, is that the ID be unique within an area for level 1 ISs and ESs, and unique within the domain for level 2 ISs". It appears that the ISO IS-ES (the ISO equivalent to ARP) is used, so that there would be no problem handling machines whose DECnet address differs from the Ethernet address. There are other features of their addressing that makes it somewhat more general than what is implied by the description here, but the document is too vague for me to be sure what it means. In particular, an entire set of addresses, using the full 8 bytes described above, is a "domain". There is some provision for connecting multiple domains, using fixed routing tables at the border nodes. The DECnet format is designed to use only 8 bytes in a fixed portion of the full ISO address, in order to facilitate hashing. There was also a comment in the talk that if necessary other address formats could be used, but handling them would be less efficient. The upshot is that I know of no problems with phase V DECnet addressing. Maybe I'll find some once I see it in detail, of course.