netnews@pikes.Colorado.EDU (Robert M. Sklar) (03/13/89)
With the growing number of machines that are part of the Internet,
and the fact that such machines have to have their name and node numbers
to be reached on the internet, I am making a proposal that a Internet group
be created.
comp.internet General Internet related discussions of interest
comp.internet.addr - simular to comp.mail.maps where a monthly posting of the
most up to date hosts table would be published and changed in Internet address
can be posted throughout the month.
Any and all feedback on this would be greatly appreciated.
Part of the reason I even suggest this is we have had problems in the delay
factor fo a university or organization changing their Internet address and
the time it takes for others to find this out, while mail, news, etc is
spooling up for days even weeks.
--
Robert M. Sklar - Network News Administrator @ CU-Denver
UUCP: {whatever}!boulder!pikes!netnews rsklar@pikes.colorado.edu
CSN: netnews@pikes.Colorado.EDU BITNET: netnews@cudenver.BITNET
***** Ignore These Four Words *****blarson@skat.usc.edu (Bob Larson) (03/13/89)
In article <2078@pikes.Colorado.EDU> netnews@pikes.Colorado.EDU (Robert M. Sklar) writes: >comp.internet General Internet related discussions of interest If you think there is a need, why don't you create it as an internet mailing list? When it succeeds, you can try convincing Erik Fair that it deservese an inet distribution newsgroup. >comp.internet.addr - simular to comp.mail.maps where a monthly posting of the >most up to date hosts table would be published and changed in Internet address >can be posted throughout the month. Definitly not needed. Hosts with reasonbly modern software use the domain name system, and have no need to bother humans with this dribble. Hosts with ancient software (that should upgrade, but havn't for some reason) should fetch HOSTS.TXT from sri-nic.arpa as frequently as they feel the need. This file only contains the 6000 or so hosts that want to be listed. You seem to have missed the required discussion period, I assume you meant "call for discussion" rather than call for votes. -- Bob Larson Arpa: Blarson@Ecla.Usc.Edu blarson@skat.usc.edu Uucp: {sdcrdcf,cit-vax}!oberon!skat!blarson Prime mailing list: info-prime-request%ais1@ecla.usc.edu oberon!ais1!info-prime-request
wisner@mailrus.cc.umich.edu (Bill Wisner) (03/13/89)
(Robert M. Sklar) >comp.internet General Internet related discussions of interest Read comp.protocols.tcp-ip, which is gatewayed with TCP-IP@SRI-NIC.ARPA. >comp.internet.addr - simular to comp.mail.maps where a monthly posting of the >most up to date hosts table would be published and changed in Internet address >can be posted throughout the month. Oh, my. "most up to date hosts table" is an oxymoron if ever there was one. Host tables have been obsolete for some time now. If you're not using nameservers for your address information you're *begging* to get hosed.
spaf@cs.purdue.edu (Gene Spafford) (03/13/89)
In article <2078@pikes.Colorado.EDU> netnews@pikes.Colorado.EDU (Robert M. Sklar) writes: > >comp.internet General Internet related discussions of interest >comp.internet.addr - simular to comp.mail.maps where a monthly posting of the >most up to date hosts table would be published and changed in Internet address >can be posted throughout the month. Gee, the NIC gave this up themselves because of the volume. There are over 60,000 sites reachable by IP on the Internet, and that changes quite a lot. Let's see, assume a really simply name and number format: foo.bar.com 129.200.92.100 That's about 30 characters per site (excluding aliases and network info) times 60,000 sites -- over 1.8Mb per month. And that isn't even addressing the fact that there is no one place that contains all that info in a form that could be posted. >Part of the reason I even suggest this is we have had problems in the delay >factor fo a university or organization changing their Internet address and >the time it takes for others to find this out, while mail, news, etc is >spooling up for days even weeks. That's a good reason to be using nameserver software. Updates are effectively instantaneous when this is done correctly. Obviously, I think the groups are unneeded and impractical. -- Gene Spafford NSF/Purdue/U of Florida Software Engineering Research Center, Dept. of Computer Sciences, Purdue University, W. Lafayette IN 47907-2004 Internet: spaf@cs.purdue.edu uucp: ...!{decwrl,gatech,ucbvax}!purdue!spaf
roy@phri.UUCP (Roy Smith) (03/13/89)
netnews@pikes.Colorado.EDU (Robert M. Sklar) writes: > I am making a proposal that a Internet group [actually 2 groups] be created. First, the manditory technical quibble -- there hasn't been any discussion about this yet, so it's premature to call for a vote. > comp.internet General Internet related discussions of interest A lot of this sort of thing already goes on in comp.protocols.tcp-ip. You could argue that this is the wrong place, but I havn't seen people screaming that talking about stuff that isn't strictly related to the tcp-ip protocol doesn't belong there. > comp.internet.addr - simular to comp.mail.maps where a monthly posting of > the most up to date hosts table would be published and changed in Internet > address can be posted throughout the month. Isn't this what the whole domain/nameserver system is all about, i.e. that nobody should ever have to have a complete, up-to-date, host table? -- Roy Smith, System Administrator Public Health Research Institute {allegra,philabs,cmcl2,rutgers,hombre}!phri!roy -or- roy@phri.nyu.edu "The connector is the network"
romkey@asylum.SF.CA.US (John Romkey) (03/14/89)
In article <2078@pikes.Colorado.EDU> netnews@pikes.Colorado.EDU (Robert M. Sklar) writes: > > With the growing number of machines that are part of the Internet, >and the fact that such machines have to have their name and node numbers >to be reached on the internet, I am making a proposal that a Internet group >be created. > >comp.internet General Internet related discussions of interest >comp.internet.addr - simular to comp.mail.maps where a monthly posting of the >most up to date hosts table would be published and changed in Internet address >can be posted throughout the month. No. This will not work, and you also should not want to do it. First off, any discussions of the Internet fall in line nicely with comp.protocols.tcp-ip. That newsgroup is gatewayed to the Internet TCP/IP mailing list. That is where most discussions about the Internet are held, except the ones by the IETF (Internet Engineering Task Force), which are generally held on IETF-specific mailing lists which are not gatewayed to the USENET. Second, there are probably >100,000 hosts on the Internet. I think you'll be unhappy about the volume of this proposed group if it succeeds (which it won't, for reasons third and fourth below). Third, there exists no central list of hosts on the Internet and no way to require new hosts to be listed on this group. That's because there already exist mechanisms which the Internet uses to keep track of such things. Fourth, only hosts on the Internet need to know the IP addresses of other hosts on the Internet to reach them. Hosts off the Internet (much of the USENET) need only the names; they address whatever it is to the appropriate name through some sort of USENET-to-Internet gateway (generally, this is only done for mail) and that gateway translates the name. The name translation is done via a distributed, replicated database which is accessed by the Domain name protocol described in various Internet RFC's. We've been doing this for several years now; if a system doesn't implement this protocol, it doesn't belong on the Internet. Here's more info on this system for those who are curious: The central host table got to be too big a long time ago; fortunately, the problem was recognized in time and the transition was made. The domains you see on the USENET (like asylum.sf.ca.us) are an echo of the domains you see on the Internet. For machines that are exclusively on the USENET, these domain names are reserved in the Internet domain namespace, but have only entries saying "forward all mail for this domain-name to this other domain-name" (MX records). The domain name database is distributed by breaking it up along the dots in names. For instance, take xx.lcs.mit.edu. There is a name server, which is replicated about five or six times (last time I checked) which knows all the domains in .edu (same for .com, .gov, .mil, .org, .net and others). It knows which computers are responsible for providing name service for .mit.edu. There are several of these computers; they know which other computers are responsible for .lcs.mit.edu. You carry this on as far as you need, and you end up with the responsiblity and burden of dealing registering host names and maintaining host tables falling on the people who own the hosts, rather than a central authority. The name server approach used by the Internet is not directly applicable to the USENET, because on the USENET you can only talk to your neighbors, making it diffcult to set up distributed servers in this fashion. If the USENET continues to grow, which I expect it to, despite rumours of its imminent death, eventually the maps which are distirbuted over comp.mail.maps will become too burdensome for many systems to deal with, and I expect that some sort of reorganization will occur in order to deal with it. One of the interesting things about the USENET (in a network sense rather than a content sense) is that these changes tend to occur in an organic, evolutionary way. Anyhow, comp.internet.addr is a bad idea and shouldn't be bothered with. The other group, comp.internet, might be useful for a more general discussion of the world wide internet of which the Internet is but a part, but this is usually covered in the various comp.mail groups, and a group with that intenet would be best named comp.worldnet or comp.matrix. -- - john romkey USENET/UUCP: romkey@asylum.sf.ca.us Internet: romkey@xx.lcs.mit.edu "Can you find me soft asylum..." - The Doors
sob@watson.bcm.tmc.edu (Stan Barber) (03/27/89)
In article <15805@oberon.USC.EDU> blarson@skat.usc.edu (Bob Larson) writes: >Hosts with ancient software (that should upgrade, but havn't for some reason) >should fetch HOSTS.TXT from sri-nic.arpa as frequently as they feel the >need. This file only contains the 6000 or so hosts that want to be >listed. WRONG. The HOSTS.TXT file contains those that MUST be listed, not those that want to be listed. New domains in the internet must have name servers. These MUST appear in HOSTS.TXT. Others do not. Most sites CANNOT have their hosts listed in HOSTS.TXT because the NIC is trying to keep the size down. Old time networks (MIT, Berkeley, Stanford) have a large number of listings because they were there first. It would be great if those sites would ask the NIC to remove unneeded host entries so that the overall size of HOSTS.TXT would be smaller and would in fact list those critical sites only. Stan internet: sob@bcm.tmc.edu Baylor College of Medicine Olan uucp: {rice,killer,hoptoad}!academ!sob Barber Opinions expressed are only mine.
john@jetson.UPMA.MD.US (John Owens) (03/28/89)
In article <1466@gazette.bcm.tmc.edu>, sob@watson.bcm.tmc.edu (Stan Barber) writes: > WRONG. The HOSTS.TXT file contains those that MUST be listed, not those > that want to be listed. New domains in the internet must have name servers. > These MUST appear in HOSTS.TXT. Others do not. Most sites CANNOT have their > hosts listed in HOSTS.TXT because the NIC is trying to keep the size down. Sorry, but nameservers have no need to be listed in HOSTS.TXT. The root nameservers will spit out information records with the nameserver addresses when responding to an NS query. A nameserver-based system can get by without ever looking at HOSTS.TXT. Essentially, HOSTS.TXT contains hosts that are there by inertia and hosts that people on (mostly) MILNET sites with old software would raise a fuss about if they could no longer reach.... [1/2 :-)] -- John Owens john@jetson.UPMA.MD.US uunet!jetson!john +1 301 249 6000 john%jetson.uucp@uunet.uu.net
louie@trantor.umd.edu (Louis A. Mamakos) (03/28/89)
If you attempt to register a name server for a second level domain (i.e. UMD.EDU), the machine in question must be registered in their database. The HOSTS.TXT file is simply an extract from that database which is used for other things as well (such as constructing the root name server zone files). Louis A. Mamakos WA3YMH Internet: louie@TRANTOR.UMD.EDU University of Maryland, Computer Science Center - Systems Programming
NU021172@NDSUVM1.BITNET (Marty Hoag) (03/29/89)
I would not favor the comp.internet.addr group as I too believe
it would give people a false sense of the hosts.txt being useful. It is
only useful in the sense of a last resort... The domain name servers
should always be used first.
However, as a "postmaster" I would like a place to find out about
known wedged systems, etc. Recently I spent some days trying to figure
out why mail was not getting to one host only to find out their software
had been on the fritz for a couple weeks. I suspect some "neighbors"
already knew that but neither hostmaster nor other sources I tried
knew what had happened.
For general host queries there is the Info-Nets list (not sure what
if any group that is gatewayed with). Is there such a group to post
announcements of outages or queries as to the health of internet systems?
Marty
-----------
Marty Hoag
ND Higher Education Computer Network US Mail: NDSU Computer Center
Phone: (701)-237-8639 PO Box 5164 / UCCS
Bitnet: NU021172@NDSUVM1 (NOTE 0 = ZERO) Fargo, ND 58105
Internet: NU021172@VM1.NoDak.EDU
UUCP: ...!uunet!ndsuvax!vm1.nodak.edu!NU021172woods@ncar.ucar.edu (Greg Woods) (03/30/89)
In article <2084NU021172@NDSUVM1> NU021172@NDSUVM1.BITNET (Marty Hoag) writes: >Is there such a group to post >announcements of outages or queries as to the health of internet systems? Yes, there is. It's called "news.config". --Greg
sob@watson.bcm.tmc.edu (Stan Barber) (04/03/89)
In article <209@jetson.UPMA.MD.US> john@jetson.UPMA.MD.US (John Owens) writes: > >Sorry, but nameservers have no need to be listed in HOSTS.TXT. The >root nameservers will spit out information records with the nameserver >addresses when responding to an NS query. A nameserver-based system >can get by without ever looking at HOSTS.TXT. > >Essentially, HOSTS.TXT contains hosts that are there by inertia and >hosts that people on (mostly) MILNET sites with old software would >raise a fuss about if they could no longer reach.... [1/2 :-)] I never said that nameservers needed to be listed for name servers to find them. If that was implied, it was bad writing on my part. What I meant to say was that the NIC only allows "important" hosts like nameservers to be listed in HOSTS.TXT. I hope I was clear in expressing that. Hosts on MILNET are not the only ones running old software unfortunately. Some sites still use 4.1BSD-style networking. :-) Imagine what will happen when sri-nic.arpa stops being a valid name or alias for the NIC computer! Stan internet: sob@bcm.tmc.edu Baylor College of Medicine Olan uucp: {rice,killer,hoptoad}!academ!sob Barber Opinions expressed are only mine.