Making Networks Self-Aware
June 1-2, 2005
Woodmark Hotel on Lake Washington
Kirkland, Washington
http://www.thewoodmark.com
Sponsored by Microsoft Research
| Home | Program Notes | Technical Program | Hotel & Travel |
Evolving a Manageable Internet
SPEAKER
Tom Anderson
University of Washington
http://www.cs.washington.edu/homes/tom/
On-demand viewing will be available
BIOGRAPHY
Tom Anderson is a Professor of Computer Science at the
University of Washington. His research concerns the practical
issues of constructing robust, secure, and efficient computer
systems, most recently focused on internetworking.
NetHealth: A Client-centric Approach to Detecting, Diagnosing and Managing Networks
SPEAKER
Victor Bahl
Microsoft Research
/~bahl
On-demand viewing will be available
BIOGRAPHY
Victor Bahl is a Senior Researcher and Manager of the Networking Research Group in
Microsoft Research. His research interests span a variety of areas in wireless
networking & mobile computing. Some of his seminal research includes: WiLIB (1997-1998), a
general purpose programming interface for wireless network cards; RADAR (1998-1999),
a signal strength based indoor user-location determination system; CHOICE (1999-2001),
a edge-server based public area wireless hot-spot network, and UCOM (2001-2003),
a multi-radio wireless system.
Towards an Internet that Never Fails
SPEAKER
Hari Balakrishnan
Massachusetts Institute of Technology
http://nms.csail.mit.edu/~hari/
On-demand viewing will be available
ABSTRACT
Suppose we want an Internet that "never fails". What exactly
does that mean, and what would we have to do to achieve that
goal? This talk will take a stab at these questions, and raise
several others in the process. Some of these will be answered
by the talk, but most of our effort will be in setting an
agenda for this exacting task.
BIOGRAPHY
Hari Balakrishnan studied at IIT Madras (B. Tech., '93) and UC Berkeley
(PhD, '98). He is now an Associate Professor of Computer Science and
Engineering at MIT. His research interests are in networked computer
systems; his recent and current projects include rcc (verifiable
Internet routing), MONET (a multi-homed overlay network for improving
network availability), IRIS (DHT protocols such as Chord and systems
such as SFR and DOA based on "flat" names), Cricket (an accurate
indoor location system, now commercially available), CarTel
(a sensor computing system for automotive applications),
Spam-I-am (spam control using quotas), and Medusa/Borealis
(data stream processing).
His honors include the ACM doctoral dissertation award, a
Sloan Foundation Fellowship, an NSF CAREER Award, MIT's
Harold Edgerton Faculty Achievement Award, seven award-winning
technical papers, and the IEEE Communication Society's Bennett
Prize.
Toward Self-directed Intrusion Detection
SPEAKER
Paul Barford
University of Wisconsin
http://www.cs.wisc.edu/~pb/
On-demand viewing will be available
ABSTRACT
Network attacks and intrusions have been a fact of life in the
Internet for many years and continue to present serious challenges for
network researchers and operators alike. The objective of our work is
to develop tools and systems that automate or otherwise enhance key
activities of network security analysts. In the first part of this
talk, I will describe our activities with dark address space
monitoring using our Internet Sink (iSink) system. iSink is a
highly scalable system that includes both
passive packet capture capability and a set of active responders that
enable details of exploits to be captured. Our results illustrate the
variability in the traffic on dark address space and the feasibility
of efficient classification and discrimination of attack types. I
will also describe recent work that uses data from dark address space
monitors to provide network "situational awareness" for security analysts.
I will also describe our recent results in the area of automated
signature generation for intrusion detection systems. We have
developed a system we call Internet Sieve (iSieve) that automatically
generates signatures from traces collected by dark address monitors.
Our evaluation of these signatures shows that they result in a much
lower false alarm rate than standard intrusion detection systems
offering the potential for much greater utility in IDS and the
possibility of truly autonomic intrusion detection. I will also
describe our current efforts at deploying both iSink and iSieve
on hardware that can be deployed throughout the Internet.
BIOGRAPHY
Paul Barford received his BS in electrical engineering from the
University of Illinois at Champaign-Urbana in 1985, and his Ph.D. in
Computer Science from Boston University in December, 2000. He is an
Assistant Professor of computer science at the University of Wisconsin
at Madison. He is the founder and director of the Wisconsin Advanced
Internet Laboratory and his research interests are in measurement,
analysis and security of wide area networked systems and network
protocols.
Self Organized Wireless LAN Mesh
SPEAKER
Christophe Diot
Intel Research Cambridge
http://cambridgeweb.cambridge.intel-research.net/people/cdiot/
On-demand viewing will be available
ABSTRACT
The meshing of wireless LANs tries to take advantage of the diversity of
access points and of the multiplicity of channels to increase the rate
obtained by end users. This study proposes a self organization scheme
that allows such a mesh to reach an optimal allocation of channels and
access points for each user. The optimality is defined in terms of delay
fairness. The self organization scheme is an avatar of the Gibbs sampler
and is fully decentralized. We describe the technological requirements,
the mathematical properties and the increase of performance of such a
scheme within the 802.11 framework.
BIOGRAPHY
Christophe Diot received a Ph.D. degree in Computer Science from INP
Grenoble in 1991. From 1993 to 1998, he was a research scientist at
INRIA Sophia Antipolis, working on new Internet architecture and
protocols. From 1998 to 2003, he created and led the IP research
group at Sprint Advanced Technology Labs. In 2003, Diot moved to
INTEL research in Cambridge, UK. His is active in the measurement
community (with work on measuring wireless networks). However, his
major interest is now on understanding how the Internet is gonna
survive mobility and wireless technologies.
Towards Reality-based Network Management
SPEAKER
Paul Francis
Cornell University
http://www.cs.cornell.edu/People/francis/
On-demand viewing will be available
ABSTRACT
Network Management is a mess: networks are managed by a hodge-podge of
tools operated by Gurus. Networks are growing in size and complexity,
including the networks in our homes, and the Guru model can't keep pace.
SNMP, the one standard tool produced by IETF (as an afterthought to the
basic architecture) can't even run until IP itself is up and running.
Recently researchers at CMU and AT&T (and other places) proposed an
exciting new fundamental architecture for network management called 4D.
4D provides a low-level routing substrate that runs immediately above the
link layer that is used to discover physical topology and allow direct
control of network equipment. We believe that 4D can be extended to also
discover and control the logical topology of the network, essentially by
discovering and controling the plumbing between drivers within hosts and
network boxes. We believe that this can be done by programming into all
drivers a small set of standard management primitives. The end result
is a network that can be understood and managed (perhaps by automated
tools) on the basis of measured reality.
BIOGRAPHY
Paul has been a researcher in computer networking for going on 20 years
now, in such organizations as MITRE, Bellcore, NTT Software Labs, and ACIRI.
Within computer networking, Paul's work has centered on routing and addressing,
with a particular liking for problems having to do with large and self-configuring
networks. Work in this vein extends from Landmark Routing, done in the late 80's,
through Yoid end-system (overlay) multicast (late 90's), to recent work on
unstructured P2P networks and more scalable end-system multicast. Notoriously,
Paul is the inventer of NAT (demonstrating great originality, if not great
prognosticative ability, judging from his bank account). Other innovations
of Paul's include shared-tree multicast, IDMaps host proximity service,
shortcut routing (through large non-broadcast subnetworks), and the
multiple-addresses approach to site multi-homing, which is the basis for
scalable routing in IPv6. Paul has recently joined the faculty at Cornell
University, where he is working on problems related to IP anycast services,
IP-level defenses against DDoS, global Internet routing, overlay multicast,
random node selection in P2P networks, the next generation of host proximity
addressing, and network management.
Self Managed Networks - Dream or Reality
SPEAKER
Jawad Khaki
Corporate Vice President, Windows Networking & Devices, Microsoft
http://www.microsoft.com/presspass/exec/jawadk/default.asp
On-demand viewing will be available
BIOGRAPHY
In his position as corporate vice president of Windows Networking & Device Technologies at Microsoft Corp., Jawad Khaki is responsible for overseeing the development of those networking technologies in Microsoft® Windows platforms. Khaki manages the group of engineers and business leaders dedicated to delivering integrated communication and device technologies that empower information workers and home users.
Since the start of his career at Microsoft in 1989, Khaki has continued to focus his passion on always-available networks for Windows-based devices that empower people with information and make it easy to seamlessly work, play and communicate.
Starting with Lan Manager, Khaki spearheaded the addition of dial-up networking, wireless networking and broadband infrastructure. He has since led the initiative to deliver information protocols, application program interfaces and core networking server components such as the Dynamic Host Configuration Protocol (DHCP), RADIUS and virtual private networking (VPN) in the Windows XP and Windows Server™ 2003 platforms. These technologies provided the foundation for Windows-based wired, wireless and peer-to-peer networking experiences for IT professionals, information workers and home users.With more than 25 years of hardware and software design experience, as well as more than 700 U.S. patents filed under his management - over 400 in the United States and more than 300 patents in other countries - Khaki brings dedication and excitement to ensuring that Windows-based PCs and devices deliver innovative, relevant and superior experiences. Khaki is responsible for determining the networking and device strategy and advances in the next generation of Windows, code-named "Longhorn."
Among Khaki's many achievements, he was appointed honorary professor by Beijing University of Post and Telecommunications in October 2003. He also contributes to his local community and was nationally recognized with the sixth annual Walter Cronkite Faith & Freedom Award by the Interfaith Alliance Foundation.
Efficient and Decentralized Discovery of Approximate Global State
SPEAKER
S. Keshav
University of Waterloo, Canada
http://www.cs.uwaterloo.ca/~keshav/
On-demand viewing will be available
ABSTRACT
The efficient computation of approximate global state lies at
the heart of several problems in massively distributed systems.
Example include routing in the Internet, sensor fusion, search
in peer-to-peer networks, and Top-K queries in stream-oriented
databases. Algorithms that determine approximate global state
enable near-optimal local decision-making with little overhead.
In this work, I will discuss some natural settings where this
problem arises, and some recent work on randomized algorithms
that navigate a four-way tradeoff between accuracy, robustness,
performance, and overhead.
BIOGRAPHY
S. Keshav is an Associate Professor and Canada Research Chair in
Tetherless Computing at the School of Computer Science, University
of Waterloo, Canada. Earlier in his career has was a researcher at
Bell Labs, an Associate Professor at Cornell, and a co-founder of
Ensim Corporation, a Silicon Valley startup. He is the author of a
widely used graduate textbook on computer networking and has been
awarded the Director's Gold Medal at IIT Delhi, the Sakrison Prize
at UC Berkeley, and the Alfred P. Sloan Fellowship. His current
interests are in infrastructural issues underlying tetherless computing.
Keshav received a B.Tech from the Indian Institute of Delhi in 1986 and
a Ph.D. from the University of California, Berkeley, in 1991.
Lessons in Engineering Self-Managed Networks
SPEAKER
Bruce Maggs
Carnegie Mellon University
http://www-2.cs.cmu.edu/~bmm/
On-demand viewing will be available
BIOGRAPHY
Bruce Maggs received the S.B., S.M., and Ph.D. degrees in computer science
from the Massachusetts Institute of Technology in 1985, 1986, and 1989,
respectively. His advisor was Charles Leiserson. After spending one year
as a Postdoctoral Associate at MIT, he worked as a Research Scientist at
NEC Research Institute in Princeton from 1990 to 1993. In 1994, he moved
to Carnegie Mellon, where he is now a Professor in the Computer Science
Department. While on a two-year leave-of-absence from Carnegie Mellon,
Maggs helped to launch Akamai Technologies, serving as its Vice President
for Research and Development, before returning to Carnegie Mellon. He
retains a part-time role at Akamai as Vice President for Research.
Maggs's research focuses on networks for parallel and distributed
computing systems. In 1986, he became the first winner (with Charles
Leiserson) of the Daniel L. Slotnick Award for Most Original Paper
at the International Conference on Parallel Processing, and in 1994
he received an NSF National Young Investigator Award. He was
co-chair of the 1993-1994 DIMACS Special Year on Massively Parallel
Computation and has served on numerous program committees iuncluding
SPAA, SODA, STOC, PODC, WWW, SIGCOMM, and IMC.
Rethinking the Systems for Network Control and Management: The Case for a New 4D Architecture
SPEAKER
David A. Maltz
Carnegie Mellon University
http://www-2.cs.cmu.edu/~dmaltz/
On-demand viewing will be available
ABSTRACT
Networks today are neither dependable nor robust infrastructure.
In this talk, I will argue that a key part of the problem is the
nature of the control and management planes that establish each
network's behavior. I will outline an alternative called the 4D
architecture that redefines the problem space into one that is
dramatically more tractable. The 4D architecture explicitly
separates network logic from distributed systems issues by
refactoring the network control system, leaving only a small
set of minimal functionality on each router. All of the decision
logic is removed from the routers and collected onto servers where
the objectives for the network can be explicitly specified and used
to directly control the network. Experimental evaluation of a
prototype shows that the architecture is feasible and enables
new capabilities missing from today's networks.
BIOGRAPHY
Dave Maltz will be joining MSR in September, where he will
continue to explore the problems of creating ubiquitous and
robust communication networks. He is currently a Postdoctoral
Fellow at Carnegie Mellon University working to simplify the
control and management of complex networks. He is a co-leader
of the 100x100 Project, which seeks to solve the problems that
stand in the way of creating networks that can deliver 100 Mbps
between all 100 Million American homes and businesses. In earlier
work, he designed the Dynamic Source Routing Protocol for ad hoc
networks and was the first employee of a start-up dedicated to
creating a metro-area wireless access network. He founded a
20-person start-up creating traffic management systems for
carrier and enterprise networks. He received his Ph.D. in
2001 from Carnegie Mellon University and his S.M. and S.B.
degrees from MIT in 1994.
The Practicality of End-User Network Monitoring
SPEAKER
Vivek Pai
Princeton University
http://www.cs.princeton.edu/~vivek/
On-demand viewing will be available
ABSTRACT
With the advent of PlanetLab, the opportunity for the average
researcher to monitor a variety of network behavior from a
number of vantage points has increased tremendously. I will
briefly discuss the experiences we have had in the following
areas: network path anomaly detection in PlanetSeer, detecting
anomalous applications in CoMon, and relating our results with
those obtained by other groups. Included in the discussion will
be where to locate such monitoring, the feasibility of data
sharing, and the utility of duplicated effort.
BIOGRAPHY
Vivek Pai is an Assistant Professor of Computer Science at
Princeton University. He works in the areas of server
performance, content distribution, and wide-area networked
systems, including anomaly detection
NetQuest: A Flexible Framework for Internet Measurement
SPEAKER
Lili Qiu
University of Texas at Austin
http://www.cs.utexas.edu/users/lili/
On-demand viewing will be available
ABSTRACT
NetQuest is a flexible framework for large-scale Internet measurement.
We apply Bayesian experimental design to design measurement
experiments that maximize the amount of information we gain
about the network path properties subject to given resource
constraints. We then apply inference techniques to reconstruct
the information of interest based on the partial, indirect
observations we get through these experiments. We further
incorporate techniques for obtaining prior information to
enhance the accuracy.
Our framework can support a variety of design requirements,
including (i) augmented design for conducting additional
experiments given existing observation, (ii) differentiated
design for providing better resolution to certain parts of
the network, and (iii) joint design for supporting multiple
users who are interested in different parts of network.
BIOGRAPHY
Lili Qiu is an Assistant Professor at University of Texas at Austin.
Before joining UT, she was a researcher at System & Networking Group in
Microsoft Research. Her research interests are wireless networks, overlay
networks, network measurement, and Web performance. She received MS and
PhD degrees in computer science from Cornell University in 1999 and
2001, respectively.
Self-Configuring Networks with a "Wafer-Thin" Control Plane
SPEAKER
Jennifer Rexford
Princeton University
http://www.cs.princeton.edu/~jrex/
On-demand viewing will be available
BIOGRAPHY
Jennifer Rexford is a Professor in the Computer Science Department at
Princeton University. Her research focuses on making data networks
like the Internet easier to design, manage, and understand. From 1996
to 2004, Jennifer worked in the IP Network Measurement and Engineering
department at AT&T Labs--Research. Along with several of her colleagues
at AT&T, Jennifer designed tools for configuration management and traffic
engineering that are in daily use in AT&T's backbone network.
In 2005, she received ACM's Grace Murray Hopper Award for her research work
on interdomain routing. Jennifer is chair of ACM SIGCOMM and is a member
of the DARPA Information Science and Technology group and the technical
advisory board of Arbor Networks. She is a senior member of the IEEE and
is coauthor of the book "Web Protocols and Practice" (Addison-Wesley, 2001).
Jennifer received her BSE degree in electrical engineering from Princeton
University in 1991, and her MSE and PhD degrees in computer science and
electrical engineering from the University of Michigan in 1993 and 1996,
respectively.
SPEAKER
Timothy Roscoe
Intel Research Berkeley
http://berkeley.intel-research.net/troscoe/
On-demand viewing will be available
BIOGRAPHY
Timothy Roscoe received a PhD from the Computer Laboratory
of the University of Cambridge, where he was a principal
designer and builder of the Nemesis operating system, as
well as working on the Wanda microkernel and Pandora multimedia
system. After three years working at an Internet startup
company in North Carolina, he worked as a researcher at Sprint's
Advanced Technology Lab in Burlingame, California, where he worked
on application hosting platforms, networking monitoring, and
assorted systems management and security problems. Mothy joined
Intel Research at Berkeley in April 2002, where his work has been
centered on PlanetLab: an open, shared platform for developing
and deploying planetary-scale services. His current research
interests include distributed query processing, network
architecture, and high-performance operating systems.
Scalable Network Proximity Estimation
SPEAKER
Puneet Sharma
Hewlett-Packard Laboratories
http://www.hpl.hp.com/personal/Puneet_Sharma/
On-demand viewing will be available
ABSTRACT
Estimation of network proximity among nodes is an important
building block in several applications like service selection
and composition, multicast tree formation, and overlay
construction. In this talk we will first discuss an enchanced
landmark-based scheme for network proximity estimation that
is scalable, accurate and robust to bad measurements.
Recently, scalable techniques have been proposed to estimate
inter-node latencies, including network coordinate systems
like GNP and Vivaldi. However, existing mechanisms for querying
such information do not scale well to a very large number of
nodes, when one wants to accurately find a set of nodes
globally closest to a given node. In the second part of
the talk we discuss research about distributing the position
data among a set of infrastructure nodes, and propose ways
of partitioning and querying this data.
BIOGRAPHY
Puneet Sharma received a Ph.D. in Computer Science from the
University of Southern California, Los Angeles in 1998. Prior
to that he earned a B.Tech. in Computer Science and Engineering
from the Indian Institute of Technology, Delhi. Currently, he is
a Research Scientist at Hewlett-Packard Laboratories, Palo Alto,
California. At HP labs he conducts research in Wireless and Mobile
Networking, Overlay Network Services, Network Measurement and
Monitioring.
Automating Network Diagnostics to Help End-Users
SPEAKER
Dave Thaler
Microsoft Corporation
/users/dthaler/
On-demand viewing will be available
ABSTRACT
Historically, network management and diagnostics have tended to focus on
helping network administrators manage networks and diagnose problems,
but end users still rely on help desks and telephone support. Our work aims
to ultimately reduce support calls and reduce the length of support calls that
to occur by having end nodes diagnose problems as much as possible,
and facilitate coordination between untrusted administrative entities in a
structured way.
BIOGRAPHY
Dave Thaler received his Ph.D. in Computer Science from the University of
Michigan in 1998, on the topic of automating network diagnostics, including
in the presence of mutually untrusted administrative entities. Dave has
been an active participant in the IETF since 1995, having now served as a
Working Group chair, a current member of the MIB Doctors group, and the
author of 15 RFCs on network management, routing, IPv6, and multicast.
At Microsoft, Dave led the effort to incorporate IPv6 into Windows XP,
and then led the effort to re-architect the entire TCP/IP stack
in Longhorn. He is now a Software Architect in the Windows Networking
division, where he works on a variety of topics including network diagnostics
and peer-to-peer protocols. He is an affiliate member of Microsoft Research.
Coordinating Chaotic Wireless Networks
SPEAKER
David Wetherall
University of Washington
http://www.cs.washington.edu/homes/djw/
On-demand viewing will be available
BIOGRAPHY
David Wetherall is an Associate Professor in the Department of Computer
Science and Engineering at the University of Washington. He joined the
faculty in 1999 after receiving his Ph.D. in computer science from MIT;
he received his B.E. in electrical engineering from the University of
Western Australia in 1989. Wetherall's thesis research pioneered active
networks, an architecture in which new network services can be introduced
rapidly using mobile code. He is broadly interested in networks, distributed
systems and operating systems, with an oerarching interest in how to best design
network protocols. Wetherall received an NSF CAREER award in 2002 and became
a Sloan Fellow in 2004.
Panel Discussion on Self-Management - What Does it Mean & Can it be Effective?
Panelist
- James Farricker,
Technical Fellow & Chief Engineer, Enterprise Network Operations, Boeing
- Craig Labovitz,
Director of Engineering & Chief Architect, Arbor Networks
- Sunjeev Pandey,
Senior Director, Global Technology Services, Microsoft Corporation
- K. Jonas Svensson,
Supportability Program Manager, Consumer Windows Support Organization,
Microsoft Corporation
BIOGRAPHYS
James Farricker is a Technical Fellow and Chief Engineer of
Boeing's Enterprise Network Organization. He is responsible for
the overall technical integrity and engineering/design activities
of the Boeing Enterprise Network, one of the largest and most complex
intranets in the US. Farricker is recognized as an expert and
innovator in the field of computing, network technologies and
data communications, with extensive experience in building large
TCP/IP Intranets, network protocols, wireless LANs, switching, and
routing technologies. He is the current technical lead of the
7E7/787 Factory Architecture Project, technical focal for Boeing
mergers and acquisitions, the lead network architect for CAS
Flight Services for the Boeing e-Enabled project, International
Space Station LAN network upgrade project, and support to Boeing
subsidiaries and airlines in the design/engineering and deployment
of worldwide computing and network infrastructure. Farricker has
been an instructor at the University of Washington extension
since 1986, where he has developed and taught a number
of technical courses in the Network Engineering, Data Communications,
Wireless Communications, Microcomputers & Networks, and Managing
Network Operations Programs. Currently, he serves on the
UW President's Visiting Committee, Network Engineering/Data
Communications, Wireless Communications, and Data and Internet
Security Advisory Boards. He has taught undergraduate and graduate
level courses in Computing Technology, Operating Systems and Data
and Computer Networking.
Craig Labovitz is Director of Engineering
and chief architect of Arbor Networks' service provider
product set. Arbor Network provides distributed network
anomaly detection and monitoring solutions to over
100 of the world's largest Internet backbones.
Before joining Arbor, Craig served as a research scientist
at Microsoft Research and Merit Network, Inc. His research
focus includes the security and fault-tolerance of large-scale
distributed systems. He is well-known for several important
early papers on Internet routing dynamics and reliability.
While at Merit, Craig served as Director of the Research
and Emerging Technologies group. His work at Merit included
design and engineering on the NSFNet backbone and Routing Arbiter
projects. Craig also served as the director of several large
National Science Foundation network architecture and routing
protocol research grants. Dr. Labovitz received his PhD. and
MSE from the University of Michigan
Sunjeev Pandey is a Senior Director in the Global Technology
Services organization at Microsoft. He is responsible for Microsoft's
Windows based IT infrastructure and services. This includes services
such as the domain controllers, DNS servers, WINS and DHCP servers,
ISA proxies and firewalls, RADIUS, VPN, and TS. Sunjeev's organization
is responsible for the OS deployment and support for all the IT
managed client machines worldwide as well as the management of
the IPSec based perimeter that Microsoft IT has deployed. His
organization is one of the key components in Microsoft IT's focus
on "dogfood" which partners with the development organizations and uses
pre-released Microsoft products to run the enterprise. He
has also worked in MSN as Group Director of Network
Engineering managing the Internet connectivity for Microsoft's
Internet presence - including network support for sites
such Microsoft.com and Windowsupdate.com
Sunjeev holds a B.S. in Computer Science and in Applied Mathematics
from the Montana College of Mineral Science and Technology.
K. Jonas Svensson is a Supportability Program Manager in the
Consumer Windows Support Organization. As an SPM, he works with
partners, end users and developers to improve the supportability
of Windows. Previous to that he was an Escalation Engineer in
Consumer Windows Support. Before joining Microsoft in 2000, Jonas
worked in technical support for Probusiness Services Inc. There he
was responsible for support and maintenance of 200 end users and
associated servers. He received a B.A. in Management of Information
Systems from Washington State University in 1995 and has obtained
several industry certifications including MCP, MCSA, CNA and CNE.
