Re: RIPE DNS Hostcount May 1993
-
To: Marten.Terpstra@localhost (Marten Terpstra)
-
From: poole@localhost (Simon Poole)
-
Date: Mon, 7 Jun 1993 19:35:37 +0200 (MET DST)
-
Cc: ripe-list@localhost, cerf@localhost
>
> I have got some major problems with pinging each and every host in the
> hostcount:
>
> - Not all hosts respond to a ping (PCs for instance) but are part of the
> Internet in my view. Given the large amounts of PCs around, this will pull
> your figure down.
>
> - Not all hosts are reachable from all parts of the Internet (AUP, firewalls,
> access filters).
Fully agreed. My reasoning for wanting this number is that it gives a
good and trustworthy lower bound on the number of network interfaces
connected (the "real" number will not be lower than this measurement).
As to the validity of the current number, you know just as good as I do
that we have no idea how "real" this is. For example a number of the
largest IP users in Switzerland are customers of ours, however since
they run seperate external DNS servers their host count is probably only
three orders of magnitude to low. On the other hand it's just so easy to
fake the numbers (want to overtake your competition? A 10 second shell
script will do the job).
The ping numbers would allow us to get some idea on how good the DNS
counts are. At least in the larger countries we can expect roughly the
same ratio of PC's/Macs/Workstations/etc. and can investigate when DNS
count/reachable hosts is totally off.
Simon
