Monday, September 25, 2006

Infrastructure Redundancy, Emergence and Coping with Disasters

Congress has recently turned its attention to net neutrality. When we speak of "net neutrality" we are generally referring to the notion of embedding preferences into the internet "pipes" to provide certain content more quickly to end users than other content. However, rarely do we question whether choosing to use only "pipes" or fiberoptic cables generates physically interconnected structures that lock in suboptimal outcomes in connectivity.

Like net neutrality, infrastructure policy regarding absence or presence of infrastructure redundancy carries with it concerns regarding market influence and information access. Our current "pipes-only" infrastructure is not the only option for structuring communication networks; by choosing not to explore alternative infrastructures, we make an inferior policy decision. Local ad-hoc networks provide more cost effective alternatives in many situations where communication is across short geographic distances. For example, two people sitting in the same room currently may need to send traffic out of the room to get packets of information across the room. With a local ad-hoc network alternative, the packets can hop directly to the neighboring machine. But, perhaps most importantly, by choosing not to experiment with alternative infrastructures, we are choosing to rely solely on our current "pipes-only" system that we know is likely to fail during disasters.

In the aftermath of Katrina, many network providers lost electrical power, communication towers or other infrastructure pieces, and internet and phone access was virtually impossible for many trapped residents of New Orleans as the water levels rose. People lost lives due to the inability to communicate. Restoring these nonfunctioning "pipes-only" networks, including simply restoring phone service, presents a time consuming and costly undertaking. Although geographically broader communication disruption did not come to pass, depending on the particular routes of fiberoptic cables, it easily could have: connectivity of other (ostensibly unimpacted) parts of the country can be meaningfully disrupted by disasters such as Katrina. This risk can and should be mitigated through infrastructure redundancy using peer-to-peer infrastructures.

During this year's DefCon, Caezar, a well-known security researcher, gave an interesting talk presenting the technological arguments for MANET networks as a viable alternative to the current "pipes-only" internet infrastructure. MANET networks (Mobile Ad-hoc Networks) are peer-to-peer networks that provide the respective peers with a method of communication and/or internet connectivity, without our current levels of reliance on "pipes." MANET networks are self-configuring networks of mobile routers and hosts connected wirelessly to form an emergent configuration or topology. The routers can move and rapidly self-organize as needed. Consequently, a MANET network may operate in either a freestanding manner or it may be connected to the internet.

It is under circumstances of crisis that peer-to-peer infrastructures such as that of MANET networks shine where a "pipes-only" approach falters -- a peer-to-peer infrastructure model can remain relatively unimpacted during disasters because the network is emergent. The nodes that connect users to each other to provide a method to communicate and obtain internet access are mobile. Each message searches out a functioning node to convey its packets until it finds one - and that connecting node may vary from message to message. The ability of a message to search for and route itself through any functioning node can mean the difference between life and death for a message's sender in natural disasters. By contrast, in a traditional "pipe" based network, if the user's one internet connection or phone network is down, connectivity self-regeneration and reorganization is not possible. The user must survive until humans fix it. As anyone who has ever had a cable or DSL problem can attest, waiting for humans to fix such problems can be a slow, painful process. In times of crisis, however, every minute matters.

Infrastructure redundancy presents a critical policy issue that warrants our attention. Our future communications regime needs to function not only in belle weather times, but also in adverse conditions of natural (or unnatural) disaster.

4 Comments:

Blogger Frank said...

Very interesting & important ideas here. I think you're especially right re health care infrastructure: developing "surge capacity" in the case of a flu epidemic or other public health disaster is absolutely necessary.

9/26/2006 8:29 AM  
Anonymous Anonymous said...

For example, two people sitting in the same room currently must send traffic out of the room to their network provider to get packets of information across a room.

Dude, your network is b0rked.

(That's a term of networking art, btw: b0rked.)

Packets around here don't leave the building unless they're addressed to an IP address outside the building. You should fire your network admin, and hire someone who knows what they're doing.

9/26/2006 9:50 AM  
Blogger Andrea Matwyshyn said...

Dear anonymous,

Thanks for the tech tweak! My statement should have said "may need to send traffic out of the room to get packets across the room...". Corrected. - AMM

9/26/2006 8:08 PM  
Blogger Dan Farber said...

This is really interesting. Of course, part of the impetus for the Internet in the first place was to decrease vulnerability to breakdown (in the form of nuclear attack, during the Cold War.) Maybe it's time to take the next step.

9/27/2006 12:37 PM  

Post a Comment

Links to this post:

Create a Link

<< Home

Google
 
Web Jurisdynamics