[98506] in North American Network Operators' Group
Re: Content Delivery Networks
daemon@ATHENA.MIT.EDU (Rodney Joffe)
Fri Aug 10 10:58:09 2007
In-Reply-To: <9b2b90b30708092255l737649e6k7861b95ba8d3a5f@mail.gmail.com>
Cc: NANOG <nanog@merit.edu>
From: Rodney Joffe <rjoffe@centergate.com>
Date: Fri, 10 Aug 2007 07:52:02 -0700
To: Paul Reubens <paulreubens11@gmail.com>
Errors-To: owner-nanog@merit.edu
On Aug 9, 2007, at 10:55 PM, Paul Reubens wrote:
> How do you engineer around enterprise and ISP recursors that don't
> honor TTL, instead caching DNS records for a week or more?
>
In my "little" bit of research and experience over the last 10 years
in this field, I have often pursued this "urban myth". It remains
largely just that.
The most common supposed violator of this was AOL. I found myself in
a position at one stage to get to the "root" of this, and was rather
impressed to find that it was indeed a myth.
We've just finished a small research project where we looked at
approximately 16 million recursive servers. The only ones violating
this were some CPE devices that ran local recursive services, and
they were generally along the lines of returning the appropriate TTL
the first time they were queried, and if the TTL was zero, they
returned a higher TTL (10000 seconds) to subsequent queries for a
short period (5 minutes). It may have been a code bug, or a designed
behavior given that these were CPE devices.
Do you have any real examples of significant recursive servers doing
this?
