MADJK@ROHVM1.BITNET (John G Kinker) (08/30/90)
We are setting up name servers and are currently trying to decide what to call various hosts. For sake of discussion assume that a network, ignoring routers, looks like: ----internet----- | hosta | hostb | | | --------subnet0---------- | | ------subnet1----- ----subnet2---- | | | | hostc hostd hoste hostf Now, if this is a business whose registered internet name is enterprise then to the internet community it is ENTERPRISE.COM. That means, I think, that the nodes might be named: hosta.subnet0.enterprise.com hostb.subnet0.enterprise.com hostc.subnet1.enterprise.com hostd.subnet1.enterprise.com hoste.subnet2.enterprise.com I also think it would be reasonable to name hosts a and b: hosta.enterprise.com hostb.enterprise.com Now the question: If all of the above is correct and valid, could one of the nodes on subnet0 be named just enterprise.com and still have all of the routing to other nodes work correctly? John G. Kinker MADJK at ROHVM1 Computer Systems Associate Rohm and Haas Company PO Box 219; Bldg 64C, Rm C303 Bristol, PA 19007 215-785-8752
oberman@rogue.llnl.gov (08/31/90)
In article <90242.124745MADJK@ROHVM1.BITNET>, MADJK@ROHVM1.BITNET (John G Kinker) writes: > Now, if this is a business whose registered internet name is enterprise > then to the internet community it is ENTERPRISE.COM. That means, I > think, that the nodes might be named: > > hosta.subnet0.enterprise.com > hostb.subnet0.enterprise.com > hostc.subnet1.enterprise.com > hostd.subnet1.enterprise.com > hoste.subnet2.enterprise.com > > I also think it would be reasonable to name hosts a and b: > > hosta.enterprise.com > hostb.enterprise.com > > Now the question: If all of the above is correct and valid, could one of > the nodes on subnet0 be named just enterprise.com and still have all of > the routing to other nodes work correctly? Time to pull up the old soapbox... NO! Naming was developed to make things easier for users, not managers. Why should users care what subnet they are on? Why should they be forced to type out long names when communicating with users on other subnets? Why should they have to keep track of who is on which subnet? Names should convey information which is useful to the user and should be as convenient to use as possible. Sub-domains for divisions in different geographic location might make a lot of sense. Or for different political bodies (such as departments or divisions). That's because such divisions provide information that is useful and memorable. And most traffic will be within the local domain and not require entering any domain information by the user. Names have NOTHING to do with network topology or addressing. (Well, I realize the problems with the inverse resolutions. But I'm on my soapbox, after all!) For anyone silly enough to want to see a more complete elucidation of my thoughts on naming, see the ESnet DECnet Naming Policy and Guidelines available from the edwg directory on icaen.llnl.gov. While these were written for a different naming system, I think the concepts are valid for the DNS. In fact, my experience with DNS led to my volunteering to write the document. R. Kevin Oberman Lawrence Livermore National Laboratory Internet: oberman@icdc.llnl.gov (415) 422-6955 Disclaimer: Don't take this too seriously. I just like to improve my typing and probably don't really know anything useful about anything.