Clean Slate Design Project

1
Clean Slate Design Project
Hui ZhangCarnegie MellonMay 2005
2
Outline

• Project overview
• Project organization management
• Status

3
Motivations for 100 x 100 Project

• Extensive concern about todays network
• Need fresh thoughts on future direction
• The upcoming deployment of national broadband
  access network
• Large investment requires full understanding of
  options and trends

4
Context

• IP was well suited to its pioneering role
  global best-effort reachability
• Global addressing scheme
• Internetworking architecture
• Simple best-effort service
• Success is a double-edged sword ? the world now
  demands more from IP and the Internet
• Converged communications services
• Dependability
• Privacy and security
• Economic sustainability
• Low cost network operations

5
Research Approaches

• Clean slate design
• What design decisions to make if we were to do it
  again?
• Holistic design
• Consider the network as a whole
• Inter-disciplinary approach architect with
  explicit considerations of economics,
  dependability, security
• Design with explicit goals of enabling tractable
  analysis and modeling

6
Why Clean Slate Design?

• A powerful research methodology that helps to
  crystallize the issues
• Smalltalk, Multics, Unix, TCP/IP
• A concrete and complete different design point
  highlights possibilities
• Understanding the target first helps to plan the
  trajectory of evolution

7
Why Clean Slate Design?

• A mind set that may result in different research
• Incremental approach to security
• How to detect and stop Blaster, Code Red?
• Clean slate design approach to security
• What would be the fundamental capability of a
  strategic adversary?
• What are the fundamental limitations/possibilities
  of any network-based or host-based security
  mechanism?
• What should be the minimal necessary set of
  layer 3 security mechanism?
• Approach used in all aspects of 100x100 research
• Backbone, access, flow control,
  control/management software, economics, security

8
Why 100x100?

• 100x100 means
• At least 100Mbps to 100 million households
• 1 Gbps to 1 million small businesses
• Exact numbers are not as important, but we would
  like to focus on a specific network
• Holistic design
• Principles derived should be general for
• Other networks such as 10 x 1000 or 1000 x 100
• Interconnection of multiple networks

9
Organic Growth vs. Clean Slate Design

• Internet is a result of organic growth
• Great for early days evolution
• But introduces protocol and network complexity,
  dependability and security concerns
• Would a clean slate design of the network be
  different from the result of organic growth?
• With clean slate design, is it possible to
  simultaneously
• simplify network and protocols and
• increase network dependability, economic
  sustainability, and security?

10
Importance of Network Structure

• Well designed networks have structures
• Access/metro tree, ring
• Interconnection butterfly topology
• Backbone full mesh
• Existing protocols not be able to take advantage
  of network structures
• Complex protocol constructs and configurations

11
The Structured Network Design Principle

• Structured network protocols taking advantage
  of network structures achieve
• Dependable, simple, and analyzable network
• Dependable, simple, and analyzable protocols

12
Technology Trends and Structured Network

• Trend 1 Deep fiber deployment key to achieving
  ubiquitous, high capacity connectivity to home
• Wireless important complementary technology to
  enable high speed, cost-effective, and ubiquitous
  access
• Copper cable may also be used for last 1000
  feet
• Fiber enables not only scalable high speed, but
  also longer distance
• Trend 2 low cost, low energy, high capacity,
  auto-configured, environmentally hardened access
  packet switches
• Trend 3 ultra high capacity (100Tbps to petabit)
  backbone switches

13
Implications for Network Architecture

• Deep fiber access switch ? large scale packet
  access networks
• In contrast, traditional access network size
  limited by copper transmission distance
• Large scale access network ultra high speed
  backbone switch ? architecture with
• regional node that terminates access network and,
• backbone network with a smaller number of richly
  connected switches

14
Project Research Organization

• Cross-Cutting Themes
• Control and Management
• Economics
• Security

• Access
• Wireless
• Fiber

Backbone
15
Project Research Components
16
Project Research Components
17
Outline

• Project overview
• Project organization management
• Status

18
Challenge of Project Coordination

• Two extreme models
• Tightly-coupled model
• How to avoid fate sharing one component failure
  should not bring down the whole project
• Loosely-coupled model
• How to avoid lack of coherency?
• How to partition the project without
  fragmentation?
• How to avoid degenerating into a committee?
• Possible to explore multiple alternatives?

19
Project Organization and Coordination

• Biweekly project management committee conference
  calls
• Weekly project meetings at each institution
• Biweekly project-wide student seminar series
• Semi-annual project retreats
• Frequent faculty/student visits, summer programs

20
Management Committee

• Members
• Hui Zhang, Dave Maltz (CMU)
• Sandy Fraser, (Fraser Research)
• Nick McKeown (Stanford)
• Ed Knightly (Rice)
• John Chuang (Berkeley)
• Larry Landweber (Internet 2)
• Activities
• Biweekly teleconference calls to discuss
  research, management, strategy
• Face-to-face meetings at major conferences
  (INFOCOM, SIGCOMM) and retreats

21
Project Meetings

• CMU weekly meetings by 100x100 and other
  interested faculty members/students
• PI and Co-PIs Zhang, Reddy, Reiter
• Other faculty members
• Marvin Sirbu, Professor of EPP, ECE, GSIA
• Bruce Maggs, Professor in CS
• Srini Seshan, Professor in CS
• Dave Farber, Professor in ISRI/SCS
• Rahul Tongia, Research Professor in ISRI/SCS

22
Student Seminar Series

• Initiated and run by graduate students
• Started in Spring 2005
• Multi-sites connected by Webex (slides),
  conference bridge (audio) , and End System
  Multicast (video)
• Once every two weeks

23
Project Retreats

• Semi-annual retreat attended by all PIs and
  students
• Dec 2003 (Pittsburgh)
• April 2004 (Berkeley)
• Dec 2004 (Pittsburgh)
• May 2005 (Pittsburgh)
• Dec 2005 (Houston, planned)

24
Outline

• Project overview
• Project organization management
• Status

25
Project Evolution

• Team started out with the shared convictions
• IP and Internet need radical re-thinking
• Clean state design should be the research
  approach
• Each institution took lead on certain core
  issues
• CMU security, network control/manageability,
  economics
• Fraser Research fiber access, network
  architecture
• Stanford backbone, regional node, congestion
  control
• Berkeley economics
• Rice wireless access

26
Where We Are

• Solid research progress on
• Backbone, fiber access, wireless, security,
  economics, network control
• Details to be covered in individual talks
• Unifying architectural themes and team synergies
  building
• Structured network e.g.
• access network, backbone network
• Holistic design, e.g.
• end-to-end lossless flow control
  (Stanford/Fraser)
• network-wide control and management
  (CMU/ATT/Stanford)
• economic informed network design
  (Berkeley/CMU/Rice/Stanford/Fraser)