craig@bbn.com (Craig Partridge) (10/10/89)
I can no longer find the original message, but someone last week asked how the number of Internet hosts is counted. The answer is that the NIC periodically runs a program that dumps the entire domain name system (takes something like 3 days to do it) and the Internet host count comes from counting the number of hosts in that dump. Note this is not a complete listing; some well-known and large sites refuse to allow transfers of their domain data. So when the NIC program locates 118,000 hosts (as it did in July) you can be sure that is an underestimate. The guess in July was that the real count was something over 150,000. By the way, there is a plan afoot to actually try to estimate the number of Internet users, by estimating the number of users per host. The IETF hopes to have that number by the end of this year... Craig
kasten@interlan.interlan.COM (Frank Kastenholz) (10/10/89)
Gee,
Seems like there is a simpler way to do it......
for (i=0; i<0xffffffff; i++)
foo = ping(i)
if (foo == answered)
number_of_hosts++;
Its only an order n problem . . . no fancy protocols, no worrying about
whether you are allowed to dump the domain tables, etc, etc...
It may take a while to finish, but that would provide a good excuse for
people to buy more/faster/bigger/<your favorite adjective here> machines.
Linearly and Inefficiently Yours...
Frank Kastenholz
Racal InterLan
(p.s. for those people who tend to believe everything they read - don't)tli@sargas.usc.edu (Tony Li) (10/11/89)
In article <8910101451.AA01170@interlan.interlan.com> kasten@interlan.interlan.COM (Frank Kastenholz) writes: Gee, Seems like there is a simpler way to do it...... for (i=0; i<0xffffffff; i++) foo = ping(i) if (foo == answered) number_of_hosts++; Its only an order n problem . . . no fancy protocols, no worrying about whether you are allowed to dump the domain tables, etc, etc... Yes, but it probably takes about 136 YEARS to run (assuming that pings take about 1 second to either respond or time out). Thanks, but we're not young enough. Tony Li - USC University Computing Services Internet: tli@usc.edu Uucp: usc!tli Bitnet: tli@gamera, tli@ramoth This is a test. This is a only a test. In the event of a real life you would have been given instructions.
michaud@devax.dec.com (Jeff Michaud) (10/12/89)
> In article <8910101451.AA01170@interlan.interlan.com> > Seems like there is a simpler way to do it...... > > for (i=0; i<0xffffffff; i++) > foo = ping(i) > if (foo == answered) > number_of_hosts++; > Yes, but it probably takes about 136 YEARS to run (assuming that pings > take about 1 second to either respond or time out). Thats only if you were to actually wait for for a response/timeout. If you just continuously sent one right after the other without waiting for the response (but sending just slow enough as to not create a traffic jam) and simply count the number of responses. I wonder how long that would take? /--------------------------------------------------------------\ |Jeff Michaud michaud@decwrl.dec.com michaud@decvax.dec.com| |DECnet-ULTRIX #include <standard/disclaimer.h> | \--------------------------------------------------------------/
meissner@twohot.rtp.dg.com (Michael Meissner) (10/16/89)
In article <5346@shlump.nac.dec.com> michaud@devax.dec.com (Jeff Michaud) writes: | > Yes, but it probably takes about 136 YEARS to run (assuming that pings | > take about 1 second to either respond or time out). | | Thats only if you were to actually wait for for a response/timeout. If | you just continuously sent one right after the other without waiting for | the response (but sending just slow enough as to not create a traffic jam) | and simply count the number of responses. I wonder how long that would | take? Note that doing this may have two adverse affects. It may cause trouble with the beancounters, for sites that pay for their links to the internet (can you say X.25?) if a large increase in the traffic occurs because of 64K pings to a class B network. Also, it will undercount some networks, which have cisco boxes (or other such gear) which are set up not do not allow inward connections except for certain hosts and/or ports. -- Michael Meissner, Data General. If compiles where much Uucp: ...!mcnc!rti!xyzzy!meissner faster, when would we Internet: meissner@dg-rtp.DG.COM have time for netnews?