Subject: Re: Slow UDP traffic
To: Vincent van Scherpenseel <firstname.lastname@example.org>
From: Steven M. Bellovin <email@example.com>
Date: 06/07/2004 13:46:42
In message <firstname.lastname@example.org>, Vincent van Sc
>On Sunday 06 June 2004 06:08, Martin Husemann wrote:
>> > Provided no packets are being lost UDP should be faster than TCP.
>> > I suspect this is one reason why NFS used TCP.
>> You mean UDP. Initially NFS used UDP, but it turns out that most of
>> the time you're better off with a TCP based implementation anyway, and
>> almost all modern NFSes do TCP.
>Jup, but DNS is still using UDP. Logical: name lookups are so frequently used
>that it really speeds up the connection (initiation).
DNS uses UDP because it's mostly simple query-response, and the 3-way
handshake to set up the connection (and the corresponding overhead to
tear it down) far exceed the cost of the actual data transfer. The
issues for NFS are far more complex.
--Steve Bellovin, http://www.research.att.com/~smb