SuperNets

1
SuperNets

• Insightful Proposal to Transform the Internet

2
The Idea

• In order to implement next-generation
  connectivity technology and applications, the
  Internet needs to become a League of SuperNets

3
The Problem

• Current Internet development is going in a
  completely different direction.
• Technology is not as much of a problem as
  politics

4
The History

• Web browsers were important in causing the start
  of the dot-com boom.
• However, the introduction and rapid upgrade by
  many to Windows 95 was much more so.

5
The Competition

• Web Services lots of industry support for
  reusable code and applications communicating
  amongst themselves.
• Expectations of 3 trillion market for Web
  Services by 2010.
• But, implementation is limited by existing
  problems with the Internet.

6
The Competition, continued

• Grid Computing Lots of computing cycles for
  sale, but who will buy them?
• Autonomic Computing Making the network smart is
  a good thing. Trouble is, its too fragile.

7
More Competition

• Embedded Sensors Put them everywhere - cheap
  too! But what are they for? The batteries die
  too fast.
• P2P Computing Peer to Peer is being used now to
  monitor realtime events affecting large amounts
  of computers. But its still not enough.

8
The Main Drawback

• All of the competing ideas are running over an
  Internet that is by design unreliable.
  Scalability, availability, security, integrity
  are limited.

9
The Wall

• Running out of IP addresses NAT helps.
• Routers too slow fast-prefix algorithms rescued
  them.
• Routers too unstable fixed with bug patches.
• Spam and Denial of Service still waiting.

10
The Current Attempts

• Overlay Networks Virtual Private Networks are an
  example. But they cant fix the underlying
  architecture of unreliable links.
• Quality of Service Mechanisms DiffServ and RSVP
  are there, but cant adequately support
  time-sensitive information transfer.

11
The League of SuperNets

• Bandwidth sharing among the Telecommunication
  companies taken to the next step - partitioning
  existing links into virtual networks that truly
  share the network capacity. The existing
  Internet could continue to run on one of these
  channels. Other channels used as dedicated and
  redundant resources.

12
The League of SuperNets

• Already some SuperNets ready to be tested
  Astrolabe (Cornell), Internet indirection
  infrastructure (i3 - UC Berkeley), RON DiffServ
  (MIT)
• Research, government, and commercial interests
  could freely test their ideas on dedicated
  SuperNets.

13
The Article

• Birman, Ken - Cornell University. The League of
  SuperNets. IEEE Internet Computing, vol. 7,
  issue 5, pp. 93-97, Sept-Oct 2003.

14
The Advertisement

• http//www.lifeaftertheoilcrash.net