map@GAAK.LCS.MIT.EDU (Michael A. Patton) (06/29/88)
How names are subdivided (with domain naming) and how networks are sudivided (with subnetting) have no intrinsic relationship. You are free to apply a relationship if that is beneficial but no relationship is required (or even implied) by the standards. In fact one of the subnets here has hosts with names in three (at least) different domains and each of those domains exist on at least half a dozen subnets. The purpose of the hierarchical domain naming system is to distribute the responsibility for assigning non-conflicting names, the purpose of subnetting is to ease the design of routing protocols internal to a single network. Mike Patton, Network Manager Laboratory for Computer Science Massachusetts Institute of Technology Disclaimer: The opinions expressed above are a figment of the phosphor on your screen and do not represent the views of MIT, LCS, or MAP.
pvm@VENERA.ISI.EDU (Paul Mockapetris) (06/30/88)
As many have pointed out there is no necessary correspondance between the name of a host, mailbox, etc. and the subnet that the host's IP address occupies. However, in the IN-ADDR.ARPA domain, which is used to map from IP addresses to host names, the name structure is constrained (by convention, not technically), with each level of the name corresponding to an octet of the IP address. Since the names break at octet boundaries, domains/zones can only break at octet boundaries, and thus if you want to separately administer address assignment, its much simpler if your subnet masks and octet boundaries correspond. For class C, this means you are out of luck. paul