Hot Interconnects 9 A Symposium on High Performance Interconnects at Stanford University on August 22 - 24, 2001


Final Program


Executive Committee

Program Committee


Call For Papers

Electronic Submission

Conference Registration

Directions / Area Links 

Sponsored by the IEEE Computer Society Technical Committee on Microprocessors and Microcomputers

Friday, August 24: Tutorials 

All information on the advance program is subject to change. 

Tutorial T1 8:30AM - 12:00PM

MPLS for Traffic Engineering

Bruce Davie, Cisco Systems, Inc.


Multiprotocol Label Switching (MPLS) has been in development for over five years and is now being widely deployed in service provider networks. One of the first applications of the technology was to introduce sophisticated traffic engineering capabilities to IP networks using constraint-based routing. This tutorial will explain how traffic engineering and constraint-based routing are implemented using MPLS, and will also explore some of the newer applications of the technology. These applications include: 

  • the control of optical devices using Generalized MPLS (GMPLS, formerly known as multiprotocol lambda switching) 
  • Diff-serv-aware traffic engineering (DS-TE), which extends the MPLS control plane to provision resources and deliver QoS guarantees in Diff-serv networks 
  • Fast Reroute, an application of contraint-based routing that enables MPLS networks to rapidly route traffic around link and node failures without the delays normally incurred by routing reconvergence


  • Introduction to MPLS: An overview of the goals of MPLS and the main principles underlying the technology, including the separation between control and forwarding planes, the label-swapping forwarding paradigm, and the use of label stacking to create levels of hierarchy. 
  • MPLS Traffic Engineering: A detailed explanation of the problems of traffic engineering in IP networks and the MPLS approach to solving them. This section will introduce constraint-based routing as a set of extensions to conventional IP routing protocols and algorithms, and will also present the signalling protocols (RSVP and CR-LDP) used for establishing traffic engineered paths. 
  • Generalized MPLS: This section will address the extensions to MPLS known as GMPLS. GMPLS will function as a unified control plane for a wide range of optical and opto-electronic switching devices, enabling optical transport networks to be managed and controlled with a standard, IP-friendly control plane. 
  • Diff-serv Aware Traffic Engineering (DS-TE): MPLS Traffic engineering was originally developed independently of any QoS technology. With the standardization of Differentiated Services (diff-serv), it is desirable to be able to perform traffic engineering independently for different QoS classes. For example, it may be desirable to send voice traffic over a significantly different path than data. This section will explain the extensions to routing protocols and to MPLS that enable per-class traffic engineering, and will show how strong QoS guarantees can be delivered using DS-TE technology. 
  • Fast Reroute: As more demanding applications move to IP networks it becomes necessary to restore service rapidly when links or nodes fail. Constraint-based routing provides a powerful tool for routing traffic around failures based on local knowledge of failures, thus enabling very fast restoration in IP networks. Several approaches to the problem, known as path, link, and node protection will be presented.

Required Background:

This talk is targeted towards network operators who are considering deployment of MPLS and towards developers who wish to gain a detailed understanding of MPLS technology. A basic level of familiarity with IP networking is expected.

Speaker Biography:

Bruce Davie works at Cisco Systems in Chelmsford, Mass., USA, where he is a Cisco Fellow. He holds a B.E. from the University of Melbourne, Australia and a Ph.D. in computer science from the University of Edinburgh. From 1988-1995 he worked at Bellcore on a variety of networking research projects. Since 1995 he has been at Cisco, where he leads a group working on the development of MPLS and quality-of-service capabilities for IP networks. He is the author of three books on networking, an active participant at the IETF, and a senior member of the IEEE.

Tutorial T2 8:30AM - 12:00PM

InfiniBand Architecture

Dhabaleswar K. Panda, Dept. of Computer Science, The Ohio State University


In recent years, there has been an explosion of commodity networking technologies supporting gigabits per second bandwidth. However, the legacy communication protocols such as TCP/IP running over these gigabit networks have been able to deliver only a fraction of the network bandwidth to the application level. Similarly, the application to application communication latency has been quite high. In order to alleviate these bottlenecks, many User-Level Networking Protocols (ULNPs) such as AM, VMMC, FM, U-Net, LAPI, and BIP have been proposed by the research community. The development of these protocols have led to the industry standard of Virtual Interface Architecture (VIA) to be available in commercial systems. Currently, the VIA standard is being implemented on many systems. While these developments are on-going, the industry has already moved ahead in drafting the next generation communication and I/O architecture standard, called InfiniBand Architecture (IBA). This new architecture eliminates the PCI bus-based architecture and supports switched connections. It incorporates high performance I/O in addition to interprocessor communication, provides QoS and protection in the network, provides multiple transport services, and allows flexibility of IBA subnets to be connected to Wide Area Networks (WANs) through routers. The IBA standard incorporates VIA functions at its verbs layer.

This tutorial is intended to provide an in-depth overview of both VIA and IBA standards. It will demonstrate how high performance communication and I/O architectures for scalable systems can be designed with this standard by taking into account interprocessor communication, network communication, and communication to I/O devices. The basic components of the IBA standard and multiple layers of this architecture will be described in detail. The new functionalities associated with this standard such as Quality of Service, Protection Domains, etc., for designing large-scale enterprise servers will be discussed. The interactions among these components and their interactions with the host, OS, Network-Interface Card (NIC), router/switch, and I/O devices to support low-latency and high bandwidth interprocessor communication and I/O will be analyzed. Open research challenges in designing scalable IBA implementations and providing support for higher-level programming models (such as distributed memory, distributed shared memory, get/put, and sockets) on these systems will be outlined. The tutorial will conclude with an overview of on-going InfiniBand and VIA related research at universities, industry, and research labs.


  • Motivation behind User-level Communication Protocols 
  • Virtual Interface Architecture (VIA) Overview 
  •  InfiniBand Architecture (IBA) Overview 
  • IBA Details
    • Addressing, Data Format, Physical layer, Link layer, Network Layer, Transport Layer, Software Transport Interface, Software Transport Verbs, Management and Services, Channel Adapters, Switches, and Routers 
  • Overview of on-going VIA and IBA Research 
  • Conclusions

Required Background:

This tutorial is intended for researchers, scientists, engineers, managers, developers, professors, and students engaged in research, design, and development of next generation communication and I/O architecture for enterprise servers as well as high-performance and scalable systems.

Speaker Biography:

Dhabaleswar K. Panda is a Professor of Computer Science at the Ohio State University. He obtained his Ph.D. in computer engineering from the University of Southern California. His research interests include parallel computer architecture, user-level communication protocols, interprocessor communication and synchronization, network-based computing, and high-performance computing. He has published over 90 papers in major journals and international conferences related to these research areas. Dr. Panda and his research group members have been doing extensive research on VIA and InfiniBand. His research group has collaborated with IBM T.J. Watson in designing a high performance VIA implementation for the IBM Netfinity cluster system. His research group is currently collaborating with Intel and IBM on VIA and InfiniBand implementations.

Dr. Panda has served on Program Committees and Organizing Committees of several parallel processing and high performance computing conferences. He was a Program Co-Chair of the 1999 International Conference on Parallel Processing, 1997 and 1998 Workshops on Communication and Architectural Support for Network-Based Parallel Computing, and a Co-Guest-Editor for two special issue volumes of Journal of Parallel and Distributed Computing on "Workstation Clusters and Network-based Computing". Currently, he is serving as a General Co-Chair for the 2001 International Conference on Parallel Processing, a Program Co-Chair of International Workshop on Communication Architecture for Clusters (CAC '01) to be held in conjunction with IPDPS '01, an Associate Editor of the IEEE Transactions on Parallel and Distributed Computing, an IEEE Distinguished Visitor Speaker, and an IEEE Chapters Tutorials Program Speaker. Dr. Panda is a recipient of the NSF Faculty Early CAREER Development Award, the Lumley Research Award at the Ohio State University (1997 and 2001), and an Ameritech Faculty Fellow Award. Dr. Panda is listed as a distinguished scientist in "Who'sWho in America" and in "American Men & Women of Science".

More details on Prof. Panda's research activities can be obtained from his CV .

More details on this tutorial can be found here.


Tutorial T3 1:30PM - 5:00PM

Bluetooth Vs. 802.11

Pravin Bhagwat, ReefEdge, Inc.


The promise of untethered computing in the workplace is becoming a reality. IEEE 802.11b, the 11Mbps wireless LAN standard, has finally arrived, and early market response has been positive.  As the WLAN market takes off, Bluetooth, another emerging standard for short-range wireless networking, is also gathering force. Several vendors have demonstrated Bluetooth products, including cordless headsets, PCMCIA cards, and LAN access points. Both standards are competing for the same airwaves, but are they also chasing the same market? Will Bluetooth and 802.11b complement each other, or will one technology eventually displace the other?


  • Review of basic concepts (RF, signal processing) and technology trends (low cost, low power, small form factor) 
  • Overview of Bluetooth 1.1 specifications
  • Overview of 802.11b specifications
  • Cost, form factor, power consumption, and co-existence of the two technologies
  • Future directions and open issues

Required Background:

The tutorial is intended for researchers and practitioners who want to learn more about Bluetooth 1.1 and 802.11b standards. The tutorial will illustrate in what ways the two technologies are similar or different. Computer professionals who want to develop better understanding of technology trends and identify new market opportunities in the space of short range wireless networking will also benefit from this tutorial. Basic understanding of layered network architecture is expected. No background in analog radio, signal processing, or wireless communication is required. Researchers who want to identify open research problems in the area of personal area networking will also find this tutorial very useful.

Speaker Biography:

Pravin Bhagwat is the principal architect at Reefedge, Inc., a networking infrastructure and software company which is building solutions for the enterprise customers to allow users to move freely throughout an in-building Bluetooth and 802.11 networks. He co-chaired the first Internet Engineering Task Force’s BOF on IP over Bluetooth. Prior to joining ReefEdge, he worked as technology consultant in the Networking Research group at AT&T Labs-Research, and as a member of research staff at IBM Thomas J. Watson Research Center. He is the chief architect of BlueSky, an indoor wireless networking system for palmtop computers, and the co-inventor of TCP splicing, a technique for building fast application layer proxies. He actively serves on program committees of mobile computing and networking conferences and has published numerous technical papers in the area of mobile computing and networking. He received his Ph.D. in computer science from the University of Maryland, College Park.

Tutorial T4 1:30PM - 5:00PM

Control and Management of Modern Optical Networks

Debanjan Saha and Sudipta Sengupta, Tellium, Inc.


The Internet transport infrastructure is moving towards a model of high-speed routers and multi-service provisioning platforms (MSPP) interconnected via an intelligent optical network. The optical network consists of optical switches connected to each other with point-to-point DWDM links in a mesh configuration. A consensus is emerging in the industry on utilizing IP-centric control plane within optical networks to support dynamic provisioning and restoration of lightpaths. Specifically, it is believed that IP routing protocols and Multi-Protocol Label Switching (MPLS) signaling protocols could be adapted for optical networking needs. At the same time, there are divergent views on how IP routers must interact with optical core networks to achieve end-to-end connectivity. This tutorial will focus on modern optical transport network architectures, network elements involved in switching and transporting optical signals, and control and management of optical networks. It will cover the following topics.

  • Optical Networking Technology & Architecture: End-to-end network model, DWDMs, Optical cross-connects, OEO vs. OOO. Multi-service Provisioning Platform (MSPP). Desired optical layer capabilities, Dynamic light-path provisioning, Fast optical layer restoration. Increased Intelligence of Optical Equipment, Migration of time-critical operations from Management-plane to Control-plane, Gradual replacement of Rings with Mesh topology.

  • Control Plane Issues: Requirements: Distributed and fast light-path provisioning, Neighbor discovery and topology dissemination, Survivability at different layers, Connectivity between different optical domains (NNI), Connectivity between optical core and client networks (UNI). Potential Solutions: Manual provisioning, NMS-based network provisioning, SONET layer bit-oriented signaling, IP-based control protocols.

  • Lightpath Routing Protocols: Similarity with IP routing protocols: Distributed and decentralized route computation. Differences: Light-paths are circuit switched. Resource information like bandwidth and SRLG required for route diverse computation of primary and backup paths. Solution involves: (i) Protocol to perform neighbor discovery: exchange detailed interface information with neighbors (OIF UNI neighbor discovery, IETF LMP), and (ii) Aggregated topology and state dissemination via appropriately modified intra-domain routing protocol (OSPF, IS-IS). Status of standardization activities (IETF, OIF).

  • Signaling Protocols: Usage: Signal lightpath set-up within the optical network (between optical switches), Signal lightpath creation request from client networks (IP routers, ATM switches, ADMs), Restoration signaling for restoring lightpaths at optical layer. Protocols: GMPLS based control plane, Re-use of MPLS signaling protocols like CR-LDP and/or RSVP-TE within the optical network, Optical User Network Interface (O-UNI, defined within OIF and IETF) for lightpath signaling between client networks (e.g. IP routers, ADM etc.) and the optical network. Status of standardization activities (IETF, OIF).

  • Restoration Protocols and Performance: Requirements, Types of restoration (e.g., dedicated/shared, span/path), Routing of primary and backup paths, Restoration signaling, Restoration capacity and restoration latency performance evaluation.

  • The "Big Picture": Emerging services: Dynamic network reconfiguration in response to changing traffic, Bandwidth trading, Advanced applications like infrastructure swapping and multi-path/multi-carrier restoration, Optical Virtual Private Networks (O-VPNs). Comparative evaluation of alternative network architectures.

Required Background:

  • General audience in the optical networking industry involved in the design and implementation of protocols, architectures, and algorithms for dynamic provisioning and restoration in optical networks. 
  • Researchers/Students from academia and research labs interested in pursuing advanced research in the optical networking area and would like to gain an understanding of the fundamental state-of-the-art concepts in this area.

Speaker Biography:

Debanjan Saha is currently with Tellium, where he is the principal architect of Tellium's network control and service management platform. Before joining Tellium, he was with IBM Research and Bell Labs Research, where he worked on various projects including high-performance switch routers, QoS in the internet, and network security. He is actively involved with various standards bodies, serves as editor of international journals and magazines, and on technical committees of national and international conferences. He has authored numerous technical articles on various networking topics, and is a frequent speaker at academic and industrial events. He received M.S. and Ph.D. degrees from the University of Maryland at College Park, and B.Tech. degree from IIT, India, all in Computer Science.

Sudipta Sengupta is currently at Tellium, where he works on the design of algorithms and protocols for dynamic provisioning and restoration in optical networks. Prior to this, he was at Oracle Corp. where he was involved in the design and development of the wireless networking platform for Oracle Mobile Applications. Prior to that, he was at Bell Labs Research, Lucent Technologies, where he worked on QoS routing and traffic engineering under the MPLS framework. He holds an M.S. degree from MIT, Cambridge, USA and a B.Tech. degree from IIT-Kanpur, India, both in Computer Science. He received the President of India Gold Medal at IIT-Kanpur for academic excellence.

For more information mail

Please send any web page comments to