[gopher] bandwidth overhead for gopher

[Cameron Kaiser]

> assuming a set selector of 10 characters long, how many bytes would that
> be........... and could you compare that to the closest example of http
> and www with a 10 character long site adress? and how many bytes would the
> http version run into?

The primary difference is Gopher has no headers and a trivial request method,
and no in-band encoding such as chunked encoding and the like. However, while
Gopher will always save you something, it may not always save you very much.
The amount of savings depends on the frequency of requests and resource size. 

For large single prolonged transfers, such as big files, the proportion of
headers to data is so small as to be negligible for especially big file sizes.
Gopher's only advantage here may well be to retard malicious discovery and
deep linking (which is why I keep my file archives generally in gopher).

For multiple short transfers, especially repeatedly, Gopher has substantially
less overhead compared to typical requests made by most current Web clients.
If you compare

/

with

GET / HTTP/1.0
Connection: close
Host: foo.bar.bletch
User-Agent: Mozilla/5.0 (6510; I; Commodore 64 BASIC 2.0)
Accept: */*

HTTP/1.0 200 OK
Server: Hello-my-name-is-Dan-and-I-will-be-your-waiter/2.0
Date: Thu, 22 Jan 2015 05:48:46 GMT
Content-Type: text/plain

you can see the ratio. If the file is short, the headers may well be a
substantial fraction of the total data sent.

-- 
------------------------------------ personal: http://www.cameronkaiser.com/ --
  Cameron Kaiser * Floodgap Systems * www.floodgap.com * ckaiser at floodgap.com
-- Atheism is a non-prophet organization. -------------------------------------