[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RRG] Consensus? Users will not adopt solutions which result in "split-system" functionality
- To: "Robin Whittle" <email@example.com>
- Subject: Re: [RRG] Consensus? Users will not adopt solutions which result in "split-system" functionality
- From: "William Herrin" <firstname.lastname@example.org>
- Date: Tue, 27 May 2008 00:11:07 -0400
- Cc: "Routing Research Group" <email@example.com>
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:sender:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references:x-google-sender-auth; b=ECPAatrZngrIOpz+F6TSHqij7Oy3LithTrdD3YPmSIuJJ77T8k5kmc323pH1ofifBoz0ntpKqYnSyCQc/9E95I/pAdZdVMtS8FHxT8Jz12L7RrOfOdMRrkT0zFL68QCDLiT5xq51J5/2ebyLcNO8xcoc3fbPqtKik5KAjoIT3GE=
- In-reply-to: <483931A5.firstname.lastname@example.org>
- References: <483931A5.email@example.com>
On Sun, May 25, 2008 at 5:30 AM, Robin Whittle <firstname.lastname@example.org> wrote:
> 1 - We have achieved consensus on very little:
Too true! So, let me turn the question on it's head:
1. What are the important, answerable questions for which we have
failed to find consensus? In other words, what important questions are
we still arguing about which have a factual answer but we don't have a
consensus on what that answer is?
2. Why is each question important?
3. How can we go about researching each question in order to find the answer?
Question: Is it permissible to have a transmission delay on the first
packet, as seen in TRRP? This has a factual answer. There is some
number of milliseconds beneath which the user never notices the delay.
There is some number of seconds above which the user is consistently
disappointed by the network performance. And there is a gray zone
between the two. We can design an experiment which lets us determine
what those numbers are.
Why is it important: If a first-packet delay is unacceptable, this
excludes virtually every pull-cache mapping system from consideration,
including TRRP and one of the LISPs.
How do we test: Build a router that simulates the delay
characteristics expected of TRRP. Install it in a den of net geeks on
their Internet link. Observe.
Simulation characteristic: If no packets for this packet's destination
have been transmitted in the past 15 minutes, delay this packet by X
ms. Otherwise, transmit immediately.
Experiment methodology: Instruct the users to inform you if they
believe the network is running slow. Have the software pick a new
delay each day. Each day there is a 50% chance that X=0. If it's not
zero then pick a random value between 50 and 2000 milliseconds. Don't
look at the numbers until after the test is over (double-blind). At
the end of the test, compute the correlation between user complaints
and the delay values that the machine logged.
William D. Herrin ................ email@example.com firstname.lastname@example.org
3005 Crane Dr. ...................... Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004
to unsubscribe send a message to email@example.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg