-----Original Message-----
From: Franck Martin [mailto:franck(_at_)sopac(_dot_)org]
Sent: Wednesday, September 12, 2001 7:44 AM
To: ISTF Participants; ietf(_at_)ietf(_dot_)org
Cc: pignet(_at_)lyris(_dot_)spc(_dot_)int; editor(_at_)isoc(_dot_)org
Subject: ISTF/IETF Work to do...? Disaster Management.
From what I read from various lists, it seems there are Internet lessons
to learn...and solutions to find. It is a trying time for the Internet
community but also a great opportunity to rise from the ashes for the
benefit off all mankind.
First let's go back to the topic of these mailing lists IETF/ISTF:
Internet. So please lets stop the SPAM and act professionaly, we will
mourn later, there a task at hand.
I'm posting to both IETF and ISTF as it relates to both part of the
Internet Society.
From what I have read so far, you can see the result of the incident on
various networks by looking at the site:
http://www.internettrafficreport.com/
You can also use the MTR tool to gather link data
http://www.bitwizard.nl/mtr/
I would suggest to all to register your router on internettrafficreport,
great site! I think if this site could also monitor long distance links
between countries it would be even better....
Patrice Husson from Luxembourg
(Patrice(_dot_)Husson(_at_)cec(_dot_)eu(_dot_)int), reported
that the transatlantic links failed from Luxembourg and that people
reverted to broadcast medium to get the news. Tomson Eric reported the
same but from Belgium. We had the same problems from the Pacific
Islands.
This is very informative on what is happening when a disaster strikes. I
think we need someone from IETF to analyse the data and produce a report
explaining what has happened on the network. Maybe someone from the ISTF
will be able to corborate the facts with some human behavior, like Dan
Kolis report that people tend to use tools they are familiar with in
emergency situations.
There has been many disaster happening in the past, like in Turkey, or
like in Taiwan earthquake where a submarine cable was cut. I think it is
time that the Internet become serious and reliable and that the IETF
work on internet and disaster to ensure absolute reliability for
emergency services which was not the case for the Vanderbilt Medical
Center for example. Can you trust your life on the Internet succesfully
delivering a piece of information or at least telling you that it was
successfully delivered or not?
I think the ISOC should include in its next Inet meeting a track on
Internet and disasters, as they did summarly in 2000. ISTF and IETF
would report on initiatives:
Can someone from the IETF lists all the work that are done or on
progress to improove the Internet when disasters strike (on an
engeenering point of view).
Can someone from the ISTF lists all the work that are done or in
progress to develop emergency network and emergency response sites
(Patrice Husson gave a few pointers).
On a last part, I remember to have a talk one year ago with Christian
Huitema, about the fact that IIS4.0 perform badly in poor network
connections. The default is that you have 2 minutes to download an ASP
page (This has changed I think with IIS5.0). I notice also the
difference between hotmail and yahoo mail which performs better in poor
network connections... Don't take me wrong I do not want this to turn
into another rant against Microsoft, I found other cases of poorly
configured Unix sites, where I talked to the administrators to fix it,
for instance many sites are not able to deal with MTU different that
1500.
Basically the IETF could come up with a testing standard that will check
that Internet interfaces and services are able to work in poor network
conditions. Usually vendors will test performance (how many transactions
per seconds) but will not test reliability, if there is heavy packet
loss, long delays, and overload. Systems need to be forgiving rather
than fast... CNN discovered it by reducing the wizzes and bangs on its
pages to reduce the bandwidth load, but did they had the best web server
to deal with the load?
Which manufacturer will dare to get its equipment or software pass the
various test and be on a compatibility list? I asked the question before
to some IETF members, and I understand that IETF is worried about giving
an advice on the quality of equipment out there in fear to loose support
from these manufacturers. But I think there are smart ways of getting
some equipment certified without the supplier being singled out.
There are many protocols that are not efficient. Take for instance
HTTP1.0 compared to HTTP1.1 which now allows to send compressed page of
HTML, FTP that allows to resume download now, as well as HTTP, only SMTP
is still not efficient, binary data as to be converted to 7bits most of
the time, and large e-mails cannot be resumed. What about all these
instant messaging systems that do not operate between each others and
have fancy protocol specs, which can be used to carry DDOS.
In many case it will help when disaster strikes but also it will help to
bridge the digital divide in countries where networks are not that
good...
Jon Postel said: "TCP implementation should follow a general principle
of robustness: be conservative in what you do, be liberal in what you
accept from others"
Finally from the inventor of the Web, not Tim the other one:
"Data is not information, information is not knowledge, knowledge is not
wisdom and none of them is action"
So action, please...
Franck Martin
franck(_at_)sopac(_dot_)org
From Patrice Husson:
----------
As far as the Internet community is concerned, it should mobilize to
improve
emergency communication in defining robust architecture capable of
handling
it working on all communication channels still operational (not limited
to
IP). It includes work on priority packeting, alternative routing,
multicasting, mobile hosts, etc. Several initiatives have started in
different regions and some are listed below. ISTF could decide for a
plan
capable of providing emergency communication at a world level.
http://perso.respublica.fr/pompiers_clic/grand_public/numero_telephone_m
onde/
http://www.telematica.de/cgalies/
http://www.nena9-1-1.org/
----------------
Michael R. Burks [mailto:mburks952(_at_)worldnet(_dot_)att(_dot_)net] reported:
----------------
Monitoring where I have seen showed a ten to twenty percent reduction in
traffic during the time people were glued to their tv sets....However
lots
of sites including the Red Cross site were unreachable.
Reports of many news sites being jammed as well. Many norm commercial
newsletter I recieve on the Internet pulled their usual content to
report on
the situation I assume to help alleviate the congestions at certain
sites.
----------------
TOMSON ERIC [mailto:Eric(_dot_)Tomson(_at_)siemens(_dot_)atea(_dot_)be]
reported:
-----------
We've been blocked for some (long) minutes, here in Belgium, in the
hours
that followed The Events.
Many companies, many users were unable to access the Web.
I know of a major direct backbone connection in a near ISP (tens of
Mbps)
that didn't respond anymore.
What could have been the reason? Too much sudden traffic? Some strategic
nodes destroyed? Bad re-routing?
-----------------
Dan Kolis <dank(_at_)hq(_dot_)lindsayelec(_dot_)com> reported:
--------
The useful Internet traffic report which sort of graphs ping like info
including packet loss, etc shows some network congestion around 19:30
GMT (0
Zulu) about 4 hours wide
....
The IETF sub commitee's on emergency traffic... I'd suggest emergency
restoration is the esssence of the goal as opposed to selective routing.
I
noted CNN for example went to text only for the heavyest portions; (or
tiny
graphics), undoubtably thinking of their bandwidth. Deciding on what
constitutes emergency traffic is very subjective, anyway. This problem
occurs in phone switches (circuit switched) too.
(1) You might use a familiar tool like a chat server normally used for
recreation to inform or inquire over a life or death situation, as
opposed
to a less familiar, never before used resource.
I noted in the media those aboard doomed aircraft using cellulars
frequently
called a named person; (husband, etc) not a emergency service. This may
be
non-rational, but it fite the scenerio (1) above.
------------------------
Bobby(_dot_)Addison(_at_)mcmail(_dot_)vanderbilt(_dot_)edu reported
-----------------
I work for Vanderbilt Medical Center in
Nashville, TN. We, like probably most hospitals on the east coast, were
sent some of the victums of this dreadful tragedy. Late yesterday
afternoon
all employees were asked to only use the net when absolutely necessary
because employees, understandbly looking for the latest news, were
consuming bandwidth needed for emergency communications.
------------------