Self-Organizing Systems

1
Self-Organizing Systems

• Emin Gün Sirer,
• Cornell University

2
Focus Area

• Self-organizing systems
• Future systems will be increasingly networked,
  increasingly complex, and increasingly difficult
  to manage manually
• Beehive / CoDoNS
• A secure peer-to-peer overlay to supplant and
  replace DNS
• MagnetOS
• A distributed operating system for ad hoc
  networks
• Herbivore
• A peer-to-peer anonymous communication system

3
Beehive

• Domain
• Critical infrastructure, DNS
• Structured peer-to-peer overlays
• Problem
• DNS is open to denial-of-service attacks
• Structured P2P systems do not provide low-latency
  lookups
• Approach

4
Beehive Approach

• general replication framework
• suitable for structured DHTs
• decentralization, self-organization, resilience
• properties
• high performance O(1) average lookup time
• scalable minimize number of replicas and reduce
  storage, bandwidth, and network load
• adaptive promptly respond to changes in
  popularity flash crowds

5
Beehive Intuition

• tunable latency
• adjust extent of replication for each object
• fundamental space-time tradeoff

0021
0112
0122
2012
6
Optimal Closed-form Solution
, 0 ? i ? K 1
xi
, K ? i ? K
1
where d b(1- ?) /?
K is determined by setting xK-1 ? 1 ?
dK-1 (K C) / (1 d dK-1) ? 1
7
Beehive Results

• Built and deployed a replacement for DNS based on
  the Beehive result
• 50-100 PlanetLab nodes additional ISPs, CNNIC
  (.cn registrar)
• high performance
• low lookup latency
• median latency of 7 ms for codons (planet-lab),
  39 ms for legacy DNS
• secure
• resilient against denial of service attacks
• load balances around hotspots
• self configures around host and network failures
  
• fast, coherent updates
• no TTLs, updates can be propagated at any time

8
MagnetOS

• Domain
• Ad hoc networks of mobile nodes
• Problem
• No programming model
• Hard to develop applications
• Need an arbiter
• Approach
• An adaptive single system image operating system

9
Approach

• Provide a unifying single-system image
  abstraction
• The entire network looks like a single machine
• MagnetOS performs automatic partitioning
• Converts applications into distributed components
  that communicate over a network
• MagnetOS provides transparent component migration
• Moves application components within the network
  to improve performance metrics

10
Automatic Partitioning

• Provide a unifying single-system image
  abstraction
• The entire network looks like a single machine
• MagnetOS performs automatic partitioning
• Converts applications into distributed components
  that communicate over a network
• MagnetOS provides transparent component migration
• Moves application components within the network
  to improve performance metrics

JVMApp
MagnetOS Rewriter
11
MagnetOS Results and Open Issues

• Power-efficient
• Extends system lifetime by a factor of four to
  five over the standard approaches
• Open issues
• Fault-tolerance for application components
• Power-efficient routing and placement techniques
• Software techniques for large-scale network
  simulations

12
Herbivore

• Domain
• Wide area networks, e.g. Internet
• Problem
• Communication protocols do not provide privacy
• Approach
• A scalable, efficient, provably anonymous
  communication system

13
Herbivore Approach

• Provably anonymous
• Dining cryptographer networks
• Scalable
• Divide and conquer the network into cliques
• Efficient
• Wire-level protocol sends only two bits per
  client

14
Herbivore Results and Open Issues

• First practical anonymous communication system
• Anonymous unicast and broadcast primitives
  layered on top of the insecure Internet
• Prototype deployed to 50 sites across the
  Internet
• Achieves 200 Kbits/second, 500ms latency!
• Design of anonymous applications still an open
  issue
• Messaging, Filesharing, Web-browsing, Voting,

15
Further Information

• E. Gün Sirer
• egs_at_cs.cornell.edu
• http//www.cs.cornell.edu/People/egs/