[dns-operations] "EarthLink Criticized for DNS Redirects"

[Edward Lewis]

At 9:59 -0500 9/15/06, John Kristoff wrote:
>On Fri, 15 Sep 2006 07:22:15 -0700
>"Barry Greene (bgreene)" <bgreene at cisco.com> wrote:
>
>>  So is this a trend? There has been talk among SPs I've been working
>
>Yeah it's a trend, the same trend as all the other attempts to limit
>the transparent flow of packet data, usually in the name of security
>and various incarnations of "inappropriate use" as defined by those
>who control the access medium.
>
>It's the continued redesign towards the mainframe/terminal architecture,
>albeit very poorly and with much more sophisticated terminals this time.
>That this sort of Internet "feature creep" happens isn't suprising, a
>bit sad in my view, because overall things seem to be degrading in as
>much as they are improving, but not surprising.

I'll add to this a less quantitative comment.

One the one hand we have a walled garden network model that is in 
place in various environments.  The walled garden is there to protect 
the economic interests of the operator - whether to make sure there 
are no external dependencies on service delivery (a "positive") or to 
lock in consumers (a cynical telling of a "negative" - from an 
engineer's point of view.)

On the other hand we have the Internet model.

Every so often I hear of a flaw in the Internet model - particularly 
its susceptibility to denial of service attacks and forged source 
addresses.  I hear this from members of the IAB, others in DNS-land. 
Each time I hear a solution to this, the solution is usually a step 
towards the "hated" walled garden approach.  Even solutions floated 
by IAB members.  (I say that to distinguish from some of the things 
said by what I'll call "economic forces" - those who are trying to 
figure out how to turn the Internet into cash making machine.  I 
don't mean this as a slight on the IAB - it' okay to "float" ideas 
for discussion.)

One thing I learned long ago is that when you have two extreme, maybe 
even polar opposite, solutions to a problem, the optimal one is going 
to be neither.  The optimal will prove to be a hybrid of the two. 
Given that, I would suspect that we will see some "walled garden" 
ingredients creep(ing) into the wide open Internet model.

I don't think it's cynically a result of economic forces taking over 
the Internet, although it could be argued that that is the "root 
cause."  I think part of this is a natural result in seeking the 
optimal approach to collectively operating the Internet.

Note that optimal doesn't mean optimal for engineering or in terms 
that are espoused in the texts used in engineering schools (like 
Tannenbaum's "Computer Networks").  Optimal here includes economic 
interests, the metrics used in that world are different from ours.

So, I too think that this "feature creep" isn't surprising.  However, 
I'm not so sure it's sad - it just represents new requirements to 
engineer towards/around.

-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                                +1-571-434-5468
NeuStar

Secrets of Success #107: Why arrive at 7am for the good parking space?
Come in at 11am while the early birds drive out to lunch.