[Air-l] "911.gov" -- Social Networks + Disaster Relief

[Robert Cannon]

Comments below.  See
www.cybertelecom.org/security/emergency.htm

--- Alexis Turner <subbies at redheadedstepchild.org>
wrote:

> On Wed, 21 Feb 2007, Frank Thomas wrote:
> 
> ::Two basic questions:
> ::1/ How will an emergency community response grid
> work when it is based 
> ::on Internet websites and the electricity supply
> breaks down ?
> 
> I've heard of a pretty amazing, advanced technology
> called a generator.  I've 
> heard of another one called a hand crank.

Generators need fuel.  ISPs had not been considered
essential services - so fuel into disaster areas
either was not permitted into the zone if it was going
to ISPs - or the fuel was rerouted to hospitals.  This
is changing, but... if you look at the assessments of
9/11 in particular, the most significant impact on the
Internet was not the 9/11 event itself, but the
running out of backup fuel a few days later.


> 
> ::2/ At which level of traffic do Internet servers
> crash because of 
> ::overload? In the old telecom times telephone
> routers in residential 
> ::areas in Europe were saturated when 8% of all
> customers in the area 

Dont compare Internet to PSTN.  They are very
different models.  PSTN assumes that only a small
percentage of subs are off hook.  If more than that go
off hook, the capacity of the switches are full and
calls are blocked.

With the Internet, in major disasters such as 9/11 and
Katrina - there was no issue of switch capacity being
overwhelmed and "calls" being blocked.  In a scenario
where the capacity of a line begins to go over 50% (I
think that is the mark), error correction will kick in
to slow down packet transmission.  This assumes that
applications are latency tolerate - which most are. 
Where the hosts back off on packet transmission, the
data still gets through but maybe a few seconds
slower.  

Conversely, as in the great baltimore train tunnel
fire and the Chinese earthquake where links are
broken, traffic is rerouted to alternative routes. 
Now these were extreme cases and it does take time for
the traffic to get rerouted, but in general in time
the traffic does get rerouted, even in the most
extreme cases, as these were.

During 9/11, the wireless telephone service switches
were overwhelmed and calls were blocked.  There are
stories of individuals implementing their companies
Y2K contingency plans going down the stairwell of the
towers using blackberry Internet type devices - the
latency tolerant distributed Internet still worked.

Final point.  The FCC Network Reliability and
Interoperability Council has always known what an
"outage" meant in terms of the PSTN.  NRIC has tried
to conceive of what an outage means for the Internet
and has largely not succeeded - alternative terms are
arising.  Instead of "outage" new terms like
"reachability" are being used.

Okay, that had to deal with the network.  For servers,
at the application layer of the network, what does it
take for a server to get overwhelmed?  How well
designed is the server??  The server host can use
dynamic routing to route traffic from server A to
mirror copies of the server anywhere in the world.
During Rita, many server farms simply moved copies of
the servers out of harms way with no impact on
performance. The server host can also slash dot the
server.  During 9/11, traffic to servers such as CNN
began to overwhelm the servers and the pipes. 
Therefore they stripped the websites of all frills -
pictures, ads, unnecessary stuff - and just hosted
text information which takes up very little bandwidth.
 Through the techniques of traffic management and web
design, servers should do well.

Final comment - then there is the issue of real time
applications such as VoIP.  The Net was not made for
real time applications.  Real time applications assume
excess capacity.  Where that excess capacity is
challenged, real time applications may struggle.

Okay, one more.  Recall that during Katrina, 1000s of
people moved from NO to Texas, got online, and were
able to access email and their servers and what they
needed.  If there is a disruption in a localized
portion of the net, a solution is to move to a
different part of the net.  Much different from the
PSTN where they could not take their PSTN phone
numbers with them.

> in New Orleans. A 
> ::close look on the living conditions of those who
> did not survive might 
> ::protect against a too strong belief in technology.


Are you suggesting that anyone believes that email is
going to stop a hurricane?  I dont really understand. 
Katrina was an extreme catastrophe.  Nevertheless, if
we look at the lessons learned, we see the importance
of communications in four phases (1) Alert of a coming
emergency (2) During the emergency, calling for help
(3) first responders and (4) restoration.  The comment
"too strong belief in technology" seems to be one from
someone who wants to set up technology as a strawman
savior, and then dismiss it.  The real quest is, how
can we improve the EAS.  How can we improve
interoperability for first responders.  How can we
improve restoring communication for survivors.  

As simple as, emergency SMS messages to phones have
become a very powerful tool not just in alerting
people, but tailoring the alert to specific areas with
specific information (where the old broadcast model
provides the same message to everyone in the broadcast
area).  CAP emergency messages are now being used for
tsunamis, earthquakes, fires, eas, local events
whatever.  

Technology is not going to stop the hurricane.  But
improved technology is helping us respond to the
event.



=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=~=
Cybertelecom :: Federal Internet Law & Policy  
            www.cybertelecom.org

          Washington hOCKEY Wiki 
           wockey.stikipad.com
News Tryouts Clubs Leagues Pickup Rinks Info