SAHARA Second Summer Retreat 4-6 June 2003 Randy H. Katz, Anthony Joseph, Ion Stoica, Doug Tygar Computer Science Division Electrical Engineering and Computer Science Department University of California, Berkeley Berkeley, CA 94720-1776 1 Retreat Goals & Technology Transfer People Project Status Work in Progress Prototype Technology Early Access to Technology Promising Directions Industrial Collaborators UC Berkeley Project Team Reality Check Friends Feedback 2 Who is Here (Industry) • Cisco – Petre Dini – Silvano Gai – David Jaffe • Crazy Tulip Systems – Chris Overton • Ericsson Research – Yuri Ismailov • Hewlett-Packard Labs – Wai-Tian Dan Tan – Mitch Trott • IBM – Pawan Goyal • KDD Japan – Ayuymu Kubota (VIF) • Microsoft Research – Venkat Padmanabhan • NEC – Yasuhiko Matsunaga (VIF) • Nortel Networks – Tal Lavian (PhD student) • NTTDoCoMo – Gang Wu • NTT MCL – Haruhiko Nishida • Sprint ATL – Gianluca Iannaccone • Univ. Helsinki – Kimmo Raatikainen (+ Students!) • Univ. NSW – Aruna Seneviratne Italics indicates Ph.D. from Berkeley VIF=Visiting Industrial Fellow 3 Green = First Retreat! Who is Here (Berkeley) • Professors – – – – Anthony Joseph Randy Katz Ion Stoica Doug Tygar – – – – Bob Miller Veronique Richard Glenda Smith Keith Sklower – – – – – Dan Adkins Sharad Agarwal Yan Chen Weidong Cui Steve Czerwinski • Technical & Admin Staff • Grad Students • Grad Students – – – – – – – – – – – – – – Paul Huang Karthik Lakshminarayanan Almadena Konrad Yin Li Sridhar Machiraju Ana Sanz Merino George Porter Anantha Rajagoplala-Rao Sonesh Surana Lakshmi Subramanian Mel Tsai Fang Yu Ben Zhao Shelley Zhuang 4 Retreat Purpose • Fourth SAHARA Retreat – Project launched 1 July 2001 – 2nd of three years: review progress, discuss “next” project – Telecomms hitting bottom! • Goal: Explore architectural elements for future networks – “Services” inside the network: code vs. protocols, location/topology-aware – Spanning: » Independent service providers » Converged data + telecomms nets » Hetero access + core nets • Co-lo w/ROC, Tygar, Jordan – Reliable, Adaptive Distributed Systems • Industrial feedback & directions – Real-world networking problems and limitations – Helping us do relevant systems research 5 Project Relationships New Net Services and Applications Adaptive, Network-Aware Applications Griffin Adaptive APIs High Level APIs Tapestry Local-aware DHT Net Storage Apps P2P Search Mobility Overlays OASIS Exposed APIs Programmable Network Elements SAHARA Composed Reachability BGP, Overlay Support Internet Indirection Infrastructure Enhanced Routing IP Layer 6 SAHARA “Elevator” Statement • New mechanisms, techniques for end-to-end services w/ desirable, predictable, enforceable properties spanning potentially distrusting service providers • Architecture for service composition and inter-operation across separate administrative domains, supporting peering and brokering, and diverse business, value-exchange, accesscontrol models 7 SAHARA Layered Reference Model for Service Composition Middleware Services End-to-End Network With Desirable Properties Enhanced “Paths” Enhanced “Links” Connectivity Plane Service Composition Applications Services Application Plane End-User Applications IP Network 8 Routing as a Composed Service • Routing as a Reachability “Service” – Implementing paths between composed service instances, e.g., “links” within an overlay network – Multi-provider environment, no centralized control • Desirable Properties – Trust: verify believability of routing advertisements – Agility: converge quickly in response to global routing changes to retain good reachability “performance” (e.g., latency)? – Reliability: detect service composition path failures quickly to enable fast recomposition to maintain reachability – Scalability and Interoperability: Adapt protocols via processing at “impedance” matching points between administrative domains 9 SAHARA Composed Services and Resource Management • Mobile Authenticated Roaming in Wireless LANs (Merino, Suzuki, Matsunaga) – Authorization control across independent administrative domains – Exploits industry-standard authentication architectures: radius, Liberty alliance – Prototype developed and deployed over last six months • Radio Resource Allocation Across Service Providers (Matsunaga) – Initial design completed – Deployment planned for Soda Hall this summer 10 SAHARA Recent Progress • Understanding “Reachability” as a Composed Network Service – BGP: Internet’s primary means for managing peering and achieving end-toend reachability » Insider’s View: Sharad Agarwal’s evaluation work with Sprint and proposed control architecture for interdomain routing – Challenge: Limited Visibility into AS Policies » Tools and Methodologies • Z. Mao: “Towards an Accurate AS-Level Traceroute Tool” • M. Caesar, L. Subramanian: “Root Cause Analysis of BGP Dynamics” – Trust Extensions » L. Subramanian: “Listen and Whisper: Security Mechanisms for BGP” • Supporting Overlays with desirable properties on IP Networks » L. Subramanian: “OverQOS: Loss-rate vs. bandwidth quality of service” » W. Cui, S. Machiraju: “Detecting Shared Bottlenecks” » Y. Chen, D. Bindel: “Tomography-Based Overlay Network Monitoring” • Ion Stoica + Students: I3, Mobility, Reliability, and P2P Networks 11 Recent MS/PhD Theses • Applications and Applications Support – Yan Chen, “Scalable Efficient Network-Aware Content Distribution Networks,” (Ph.D., expected Dec 2003) • Resource Allocation and Management – Jimmy Shih, “Applying Congestion Pricing at Access Points for Voice and Data Traffic,” (Ph.D., May 2003) – Sridhar Machiraju, “A Scalable and Robust Solution for Bandwidth Allocations,” (M.S., May 2003) • Reachability as a Service – Morley Mao, “Solving the Interdomain Routing Puzzle—Understanding Interdomain Routing Dynamics,” (Ph.D., expected December 2003) – George Porter, “Traffic Matrix Estimation for Low-loss Routing in Hybrid Networks,” (M.S., May 2003) • Reliability as a Design Metric – Bhaskar Raman, “An Architecture for Availability and Performance in Wide-Area Service Composition,” (Ph.D., December 2002) – Fang Yu, “Study of the Restoration Path Block Problem in Optical Networks,” (M.S., December 2002) – Weidong Cui, “Backup Path Allocation Based on a Correlated Link Failure Probability Model in Overlay Networks,” (M.S., May 2003) 12 Overlays and Active Services for Internetworked Storage – Programmable Network Elements – Generalized programming model for such elements – Applications of Interest » Network Services: L7 switching, firewalls, intrusion detection, storage virtualization, network monitoring and management, etc. » Wide-Area Storage, iSCSI support 13 Winter03 Retreat Feedback • Retreat Organization: – – – – Encourage more interactive discussion, fewer and/or shorter talks Provide more context, clearly define terms, review related work Place read-ahead material on the Web before retreat Organize sessions according to topic and/or research area • Technical Comments: – Identify the application for overlays, P2P, programmable networks – Tackling complexity: network and security configuration are good and hard problems – Strategy to migrate overlay technology into networking mainstream – Consider control plane (“signaling system”) as well as data plane – Streaming media as a workload/application distinct from Web – What is criteria for processing in the network vs. at the end-points? How is session state exploited in programmable networks? – More statistical/mathematical expertise needed; economic/game theory analyses could be useful 14 Plan for the Retreat • Wednesday, 4 June 2003 – 1000-1200 Drive to Santa Cruz – 1200-1300 Lunch – 1300-1500 Retreat Overview and Introductions (Randy) » Retreat Overview & Sahara Progress, Randy » I3 Status, Ion » Griffin Status, Anthony » UHelsinki Research, Kimmo – 1500-1530 Break – 1530-1655 Research Highlights (Anthony) » “Modeling, Analysis, and Prediction of Non-Stationary Network Behavior,” Almadena » “Tomography-based Overlay Network Monitoring,” Yan » “Overlayed Overlay Networks,” Mukund – 1655-1715 Break – 1715-1830 Research Highlights (Anthony) » “Detecting Shared Bottlenecks,” Weidong, Machi » “Wireless LAN Authenticated Roaming,” Ana, Yas » “Scalable Ad-Hoc Routing with Location Information,” Ananth – 1830-2000 Joint Dinner – 2000-2100 Evening Session, New Directions » Overview of “Robust Adaptive Distributed Systems,” Randy » Panel Session Katz/Patterson/Stoica/Tygar – 2100- Social Hour 15 Plan for the Retreat • Thursday, 5 June 2003 – 0730-0830 Breakfast – 0830-1000 I3 Based Overlays and P2P Networks (Ion) » “Towards a More Secure and Flexible Network Infrastructure,” Dan » “Infrastructure Primitives for Overlay Networks,” Karthik » “Fast Failure Detection in Overlay Networks,” Shelley » “Load Balancing in p2p Systems,” Sonesh – 1000-1030 Break – 1030-1200 OASIS (Randy) » “Overview of Programmable Networks White Paper” George, Mel, Li » Discussion: Apps and Services Development Principles for Programmable Networks – 1200-1630 Box Lunches and Long Break – 1630-1800 Joint RADS Breakouts » Challenges for Service/Server/Network Monitoring, Measurement & Management » Reliability Benchmarking for Networks, Servers, and Services » Managing Denial of Service and Service Failures in Systems » Deploying P2P and Overlay Networks » Minimizing the Effect of Operator Errors and Misconfigurations in System Failures » Verifying and Learning Correct Service and Protocol Behaviors » Other Topics – 1800-1930 Joint Dinner – 1930-2100 Student Posters & Social Hour – 2100- Wild Ideas & Open Mic (Armando) 16 Plan for the Retreat • Friday, 6 June 2003 – 0730-0830 Breakfast – 0830-0930 New Research Opportunity Synthesis (Randy & Dave) – 0930-1000 Break/Room Checkout/Photo Session – 1000-1200 Industrial Feedback (Randy) – 1200-1300 Lunch – 1300-1500 Drive back to Berkeley 17 Structure of the Thursday Breakouts • Grad student facilitators and note takers; no two participants from same company at same breakout! Signup in advance. – 15 minutes for group to define what topic area means to them, and to capture that discussion on one powerpoint slide – 10 minutes for participants to individually describe (up to) 3 “big” research questions in the breakout topic area – 30 minutes to post and discuss the questions thus enumerated – 20 minutes to vote for the 3 most important (everyone gets 3 votes, cumulative voting allowed!) – 15 minutes to summarize the meaning of the most important 3 questions and to capture these on a second powerpoint slide 18 Recent SAHARA-Related Publications • • • • • • • • S. Agarwal, C. Chuah, S. Bhattacharyya, C. Diot, “Impact of BGP Dynamics on Intra-domain Traffic Patterns in the Sprint Backbone,” 27th North American Network Operators' Group Meeting, NANOG 27, (February 2003). B. Raman, R. H. Katz, “Load Balancing and Stability Issues in Algorithms for Service Composition,” IEEE Infocomm Conference, San Francisco, California, (April 2003). S. Agarwal, C. N. Chuah, R. H. Katz, “OPCA: Robust Interdomain Policy Routing and Traffic Control,” Proceedings OpenArch 2003, San Francisco, CA, (April 2003). W. Tan, W. Cui, J. Apostolopoulos, “Playback Buffer Equalization for Streaming Media using Stateless Transport Prioritization,” Packet Video Workshop, (April 2003). S. Zhuang, K. Lai, I. Stoica, R. H. Katz, S. Shenker, “Host Mobility using an Internet Indirection Infrastructure,” First International Conference on Mobile Systems, Applications, and Services (ACM/USENIX Mobisys), San Francisco, CA, (May 2003). Z. Mao, J. Rexford, J. Wang, R. H. Katz, “Towards an Accurate AS-Level Traceroute Tool ,” Proceedings ACM SIGCOMM Conference, Karlsruhe, Germany, (August 2003). B. Raman, R. H. Katz, “An Architecture for Highly Available Wide-Area Service Composition,” Computer Communications Journal, Special Issue on “Recent Advances in Communication Networking”, (2003). Y. Chen, L. Qiu, W. Chen, L. Nguyen, R. H. Katz, “Efficient and Adaptive Web Replication using Content Clustering,” IEEE Journal on Selected Areas in Communications (J-SAC), Special Issue on “Internet and WWW Measurement, Mapping, and Modeling,” (2003), to appear. 19 Sahara Overview Randy H. Katz Univ. of California Berkeley, CA 94720-1776 20