book cover image High-Speed Networking:
A Systematic Approach to High-Bandwidth Low Latency Communication

James P.G. Sterbenz and Joseph D. Touch

with contributions from Julio Escobar, Rajesh Krishnan, and Chunming Qiao
technical editor A. Lyman Chapin
Wiley, 2001
LC: TK5105.5 S743


This book presents a comprehensive introduction to all aspects of high-speed networking, and is not about any particular protocol or standard, but is rather a systematic approach to the principles that guide the research and design of high-speed networks, protocols, and applications. We avoid generic terms like gigabit networking and terabit networking since these are dependent on a particular technology at a particular point in time, and also differ per link, per flow, and in aggregate. Quantifiers are applied in specific contexts and in examples as appropriate.

The network is a complex system of systems, and high-speed networking does not result from the design of individual components or protocols in isolation. Thus, this book presents a systemic approach to high-speed networks, where the goal is to provide high-bandwidth and low latency to distributed applications, and to deal with the high bandwidth-×-delay product that results from high-speed networking over long distances.


A set of fundamental axioms is presented that govern the design of all systems:

  1. Know the past, present, and future
    1. Not invented here corollary (1980s: Unix and TCP/IP)
    2. Not invented here corollary (1990s: MS-Windows and x86 PC)
  2. Application primacy
    1. Field of dreams vs. killer app dilemma
    2. Interapplication delay
    3. Network bandwidth and latency
    4. Networking importance in system design
  3. High-performance paths goal
    1. Path establishment corollary
    2. Path protection corollary
    3. Store-and-forward avoidance
    4. Blocking avoidance
    5. Contention avoidance
    6. Efficient transfer of control
    7. Path information assurance tradeoff
  4. Limiting constraints
    1. Speed of light
    2. Channel capacity
    3. Attenuation and transmission power*
    4. Switching speed
    5. Cost and feasibility
    6. Heterogeneity
    7. Policy and administration
    8. Backward compatibility inhibits radical change
    9. Standards both facilitate and impede dilemma
  5. Systemic optimisation principle
    1. Consider side effects
    2. Keep is simple and open
    3. System partitioning corollary
    4. Flexibility and workaround corollary

High-Speed Network Design Principles

The Systemic Optimisation Principle is the basis for a set of principles and corollaries that are the basis for the architecture and design of high-speed networks and enable high-speed communication between users:

  1. Selective optimisation
    1. Second-order effect corollary
    2. Critical path corollary
    3. Functional partitioning and assignment corollary
  2. Resource tradeoff principle
    1. Resource tradeoffs change
    2. Optimal resources utilisation vs. over-engineering tradeoff
    3. Support for multicast
  3. End-to-end argument
    1. Hop-by-hop performance enhancement corollary
    2. Endpoint recursion corollary
  4. Protocol layering principle
    1. Layering as an implementation technique performs poorly
    2. Redundant layer functionality corollary
    3. Layer synergy corollary
    4. Hourglass corollary
    5. Integrated layer processing (ILP) corollary
    6. Balance transparency and abstraction vs. hiding
    7. Support a variety of interface mechanisms
    8. Interrupt vs. polling
    9. Interface scalability corollary
  5. State management principle
    1. Hard state vs. soft state vs. stateless tradeoff
    2. Aggregation and reduction of state transfer
    3. Hierarchy corollary
    4. Scope of information tradeoff
    5. Assumed initial conditions
    6. Minimise control overhead
  6. Control mechanism latency principle
    1. Minimise round trips
    2. Exploit local knowledge
    3. Anticipate future state
    4. Open- vs. closed-loop control
    5. Separate control mechanisyms
  7. Distributed data principle
    1. Partitioning and structuring of data
    2. Location of data
  8. Protocol data unit principle
    1. PDU size and granularity
    2. PDU control field structure
    3. Scalability of control fields

Topic Organisation

Following the introduction (Chap. 1) and initial presentation of axioms and principles (Chap. 2) subsequent chapters of the book apply the axioms and principles to all areas of networking and communication [full table of contents | list of figures | list of tables]:

  1. Introduction
  2. Fundamentals and design principles
  3. Network architecture and topology
  4. Network control and signalling
  5. Network components
    1. Communication links
    2. Switches and routers
  6. End systems
  7. End-to-end protocols
  8. Networked applications
  9. Future directions and conclusions

A set of design techniques are also introduced in Chap.2) and applied as appropriate throughout the book:

  1. Scaling time and space
  2. Cheating and masking the speed of light
  3. Specialised hardware implementation
  4. Parallelism and pipelining
  5. Data structure optimisation
  6. Latency reduction


While the book is not about any particular protocol suites or systems in particular, examples of the application of these principles and techniques are given, including:

  1. Fundamentals examples
    1. ATM and the revenge of layered systems
    2. ATM cell size and format
  2. Network architecture and topology examples
    1. The vision and tragedy of Iridium
    2. ATM virtual connection hierarchy
    3. When is a hop not a hop?
    4. Multicast
    5. Content cache location
  3. Network control and signalling examples
    1. ATM signalling
    2. Internet and ATM service models
    3. Mobile IP and cellular telephony
    4. Active network management
  4. Network components examples
    1. ITU transport networks and GFP*
    2. SONET links
    3. OTN (optical transport network)*
    4. The evolution of Ethernet
    5. The evolution of 802.11 WiFi*
    6. SONET architecture
    7. ATM
    8. MPLS
    9. IP packet formats
    10. Network processor examples: IXP 1200/2400/2800/2850, PowerNP 4GS3, C-Port C-5*
  5. End system examples
    1. The XTP/PE interface
    2. The failure of ATM to the desktop
  6. End-to-end protocol examples
    1. Email and web caches
    2. End-to-end Internet protocols
    3. TCP for transactions (T/TCP)
    4. Persistent HTTP
    5. ATM segmentation and frame discard
    6. Error control in SNA
    7. ATM traffic classes
    8. TCP congestion control
    9. TCP optimisations for high-speed
    10. AES high-speed encryption*
  7. Networked application examples
    1. Interplanetary Web
    2. MPEG-2 compression

Target audience

Target Audience: This book is ideally suited to be used for a graduate-level special topics course in high-speed networking. It also intended for computer scientists and engineers who want to learn about high-speed systems design, and who may have expertise in a narrow aspect of high-speed networking (such as switch design), but want to gain a broader understanding of all aspects of high-speed networking and the impact that their designs have on overall network performance.

*new since publication; see additions page

Supplementary Materials

Errata and additions

Tutorial notes for instructors and students

Ordering Information

ISBN 0-471-33036-1 (hardcover)

Wiley US hardcover e-book instructor evaluation copy UK hardcover

Amazon US hardcover e-book UK hardcover e-book FR hardcover
Barnes and Noble US hardcover
Chapters CA hardcover

[ UP ]

Last updated 3 November 2003 – Valid XHTML 1.0Lynx inspectedW3C AAA Conformance
©2003 James P.G. Sterbenz <>