[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[ih] how big was the host file
Two other phenomena of the years before the DNS came into use was
that some sites would (in a sense) reserve a series of names so that, as
host computers were acquired, these hosts would have related names.
So the first phenomenon was the set of imaginative naming
conventions, famously starting with the use of plants from the Sunset
Garden book at PARC, were interesting in themselves. Names of planets
and names of characters from Leave it to Beaver were among the less
famous examples. Many members of this list can probably remember local
examples. When the net was young, a given site might be known for such
a name set, then if a new host that fit that name series came up,
people would know/suspect that the host was from the site known for that
set of names.
The second phenomenon was the reserving of names for hosts that did
not yet exist. This reserving was necessary to preserve the integrity
of the logic of a series of names. But I mention it because, when I
looked at a hosts.txt file in 1984, it seemed that the size of the file
was significantly influenced by this 'reserving'.
I mention this, in part, let anyone assume that the number of entries
in a given hosts.txt file was an accurate estimate of the number of
actually existing hosts.
On 2/5/20 16:52, Joseph Touch via Internet-history wrote:
> FWIW - the host file was only for off-site systems; it was (AFAICT typically) augmented with the local list of hosts. At some places, ihis was fairly large as well.
>> On Feb 5, 2020, at 2:15 PM, Brian E Carpenter via Internet-history <internet-history at elists.isoc.org> wrote:
>> A little Google work got me:
>> June 1985, 1528 hostnames.
>> December 1986, 4480 hostnames.
>> May 1987, 5343 hostnames.
>> November 1988, 7083 hostnames.
>> Compare these to the RFC1296 size estimates of the Internet:
>> 1985: 1961 nodes
>> 1986: 5089
>> 1987: 28174
>> 1988: 56000
>> In other words, the NIC table was overwhelmed by growth by 1986 and its size had ceased to matter by then. From those numbers, it looks as if DNS superseded hosts.txt in practice around mid-1986. Read the introductory text of RFC1296 for more.
>> Brian Carpenter
>> On 06-Feb-20 09:53, Michael Kj?rling via Internet-history wrote:
>>> On 5 Feb 2020 20:33 +0000, from internet-history at elists.isoc.org (Jacques Latour via Internet-history):
>>>> How big was the host file before the DNS came in action? 200K entries?
>>> I'm not sure, but 200K entries sounds large.
>>> RFCs 1034 and 1035 are dated November 1987. Though work began earlier,
>>> that's probably a decent approximation for when "DNS came into
>>> RFC 1296 (January 1992) provides some data points on Internet growth
>>> for the period 1981-1991. That one gives the number of hosts with an
>>> IP address on the Internet in December 1987 as 28,174.
>>> The next data point in that RFC is about half a year later, in July
>>> 1988, at 33,000; followed by October 1988, 56,000.
>>> Even taking into account that migrating to DNS probably wasn't
>>> instant, my guess for the size of the hosts file in late 1987 would be
>>> a lot closer to 20K entries than 200K.
>>> Extrapolating from the data in RFC 1296, the Internet would have
>>> passed 200K IP hosts some time in mid-1990, with some 85% of those
>>> (net) added post-DNS.
>> Internet-history mailing list
>> Internet-history at elists.isoc.org