1 CS 525 Advanced Distributed Systems Spring 09 Indranil Gupta Lecture 7 More on Epidemics (or “Tipping Point Protocols”) February 12, 2009 (gatorlog.com)

Slides:



Advertisements
Similar presentations
Quality-of-Service Routing in IP Networks Donna Ghosh, Venkatesh Sarangan, and Raj Acharya IEEE TRANSACTIONS ON MULTIMEDIA JUNE 2001.
Advertisements

CS 425 / ECE 428 Distributed Systems Fall 2014 Indranil Gupta (Indy)
Exploring Energy-Latency Tradeoffs for Broadcasts in Energy-Saving Sensor Networks AUTHOR: MATTHEW J. MILLER CIGDEM SENGUL INDRANIL GUPTA PRESENTER: WENYU.
HIERARCHY REFERENCING TIME SYNCHRONIZATION PROTOCOL Prepared by : Sunny Kr. Lohani, Roll – 16 Sem – 7, Dept. of Comp. Sc. & Engg.
A Presentation by: Noman Shahreyar
5/2/2015 Wireless Sensor Networks COE 499 Sleep-based Topology Control II Tarek Sheltami KFUPM CCSE COE
Practical Exploitation of the Energy-Latency Tradeoff for Sensor Network Broadcast Matthew J. Miller (Cisco Systems) Indranil Gupta (Univ. of Illinois-Urbana)
Optimizing Buffer Management for Reliable Multicast Zhen Xiao AT&T Labs – Research Joint work with Ken Birman and Robbert van Renesse.
Rumor Routing Algorithm For sensor Networks David Braginsky, Computer Science Department, UCLA Presented By: Yaohua Zhu CS691 Spring 2003.
Epidemic Techniques Algorithms and Implementations.
Gossip Algorithms and Implementing a Cluster/Grid Information service MsSys Course Amar Lior and Barak Amnon.
Feb 7, 2001CSCI {4,6}900: Ubiquitous Computing1 Announcements.
1 Routing Techniques in Wireless Sensor networks: A Survey.
Lecture 7 Data distribution Epidemic protocols. EECE 411: Design of Distributed Software Applications Epidemic algorithms: Basic Idea Idea Update operations.
Gossip algorithms : “infect forever” dynamics Low-level objectives: – One-to-all: Disseminate rumor from source node to all nodes of network – All-to-all:
Reliable Group Communication Quanzeng You & Haoliang Wang.
Gossip Scheduling for Periodic Streams in Ad-hoc WSNs Ercan Ucan, Nathanael Thompson, Indranil Gupta Department of Computer Science University of Illinois.
Monday, June 01, 2015 ARRIVE: Algorithm for Robust Routing in Volatile Environments 1 NEST Retreat, Lake Tahoe, June
Receiver-driven Layered Multicast S. McCanne, V. Jacobsen and M. Vetterli SIGCOMM 1996.
1 LINK STATE PROTOCOLS (contents) Disadvantages of the distance vector protocols Link state protocols Why is a link state protocol better?
Multiple constraints QoS Routing Given: - a (real time) connection request with specified QoS requirements (e.g., Bdw, Delay, Jitter, packet loss, path.
1 Ultra-Low Duty Cycle MAC with Scheduled Channel Polling Wei Ye Fabio Silva John Heidemann Presented by: Ronak Bhuta Date: 4 th December 2007.
Adaptive Self-Configuring Sensor Network Topologies ns-2 simulation & performance analysis Zhenghua Fu Ben Greenstein Petros Zerfos.
Epidemic Techniques Chiu Wah So (Kelvin). Database Replication Why do we replicate database? – Low latency – High availability To achieve strong (sequential)
MAC Layer Protocols for Sensor Networks Leonardo Leiria Fernandes.
Multicast Communication Multicast is the delivery of a message to a group of receivers simultaneously in a single transmission from the source – The source.
EPIDEMIC TECHNIQUES Ki Suh Lee. OUTLINE Epidemic Protocol Epidemic Algorithms for Replicated Database Maintenance Astrolabe: A Robust and scalable technology.
Presentation Title Subtitle Author Copyright © 2002 OPNET Technologies, Inc. TM Introduction to IP and Routing.
Distributed Quality-of-Service Routing of Best Constrained Shortest Paths. Abdelhamid MELLOUK, Said HOCEINI, Farid BAGUENINE, Mustapha CHEURFA Computers.
Epidemic Algorithms for replicated Database maintenance Alan Demers et al Xerox Palo Alto Research Center, PODC 87 Presented by: Harshit Dokania.
Communication (II) Chapter 4
1 6.4 Distribution Protocols Different ways of propagating/distributing updates to replicas, independent of the consistency model. First design issue.
Itrat Rasool Quadri ST ID COE-543 Wireless and Mobile Networks
Probabilistic Broadcast Presented by Keren Censor 1.
Tanenbaum & Van Steen, Distributed Systems: Principles and Paradigms, 2e, (c) 2007 Prentice-Hall, Inc. All rights reserved DISTRIBUTED SYSTEMS.
Epidemics Michael Ford Simon Krueger 1. IT’S JUST LIKE TELEPHONE! 2.
A Randomized Error Recovery Algorithm for Reliable Multicast Zhen Xiao Ken Birman AT&T Labs – Research Cornell University.
LECTURE9 NET301. DYNAMIC MAC PROTOCOL: CONTENTION PROTOCOL Carrier Sense Multiple Access (CSMA): A protocol in which a node verifies the absence of other.
Power Save Mechanisms for Multi-Hop Wireless Networks Matthew J. Miller and Nitin H. Vaidya University of Illinois at Urbana-Champaign BROADNETS October.
CSE 486/586, Spring 2013 CSE 486/586 Distributed Systems Gossiping Steve Ko Computer Sciences and Engineering University at Buffalo.
Group 3 Sandeep Chinni Arif Khan Venkat Rajiv. Delay Tolerant Networks Path from source to destination is not present at any single point in time. Combining.
Co-Grid: an Efficient Coverage Maintenance Protocol for Distributed Sensor Networks Guoliang Xing; Chenyang Lu; Robert Pless; Joseph A. O ’ Sullivan Department.
Load-Balancing Routing in Multichannel Hybrid Wireless Networks With Single Network Interface So, J.; Vaidya, N. H.; Vehicular Technology, IEEE Transactions.
GPSR: Greedy Perimeter Stateless Routing for Wireless Networks EECS 600 Advanced Network Research, Spring 2005 Shudong Jin February 14, 2005.
Data Collection and Dissemination. Learning Objectives Understand Trickle – an data dissemination protocol for WSNs Understand data collection protocols.
2007/1/15http:// Lightweight Probabilistic Broadcast M2 Tatsuya Shirai M1 Dai Saito.
KAIS T High-throughput multicast routing metrics in wireless mesh networks Sabyasachi Roy, Dimitrios Koutsonikolas, Saumitra Das, and Y. Charlie Hu ICDCS.
1 Gossip-Based Ad Hoc Routing Zygmunt J. Haas, Joseph Halpern, LiLi Cornell University Presented By Charuka Silva.
An Energy Efficient MAC Protocol for Wireless LANs, E.-S. Jung and N.H. Vaidya, INFOCOM 2002, June 2002 吳豐州.
Tufts Wireless Laboratory School Of Engineering Tufts University Paper Review “An Energy Efficient Multipath Routing Protocol for Wireless Sensor Networks”,
a/b/g Networks Routing Herbert Rubens Slides taken from UIUC Wireless Networking Group.
SMAC: An Energy-efficient MAC Protocol for Wireless Networks
1 An Adaptive Energy-Efficient MAC Protocol for Wireless Sensor Networks Tijs van Dam, Koen Langendoen In ACM SenSys /1/2005 Hong-Shi Wang.
An Energy-Efficient MAC Protocol for Wireless Sensor Networks Speaker: hsiwei Wei Ye, John Heidemann and Deborah Estrin. IEEE INFOCOM 2002 Page
Evaluation of ad hoc routing over a channel switching MAC protocol Ethan Phelps-Goodman Lillie Kittredge.
CS541 Advanced Networking 1 Contention-based MAC Protocol for Wireless Sensor Networks Neil Tang 4/20/2009.
Networks, Part 2 March 7, Networks End to End Layer  Build upon unreliable Network Layer  As needed, compensate for latency, ordering, data.
Exploring the Energy-Latency Trade-off for Broadcasts in Energy-Saving Sensor Networks Matthew J. Miller, Cigdem Sengul, Indranil Gupta Department of Computer.
Oregon Graduate Institute1 Sensor and energy-efficient networking CSE 525: Advanced Networking Computer Science and Engineering Department Winter 2004.
CSE 486/586 CSE 486/586 Distributed Systems Gossiping Steve Ko Computer Sciences and Engineering University at Buffalo.
Sensor Networks Katia Obraczka Winter 2005 MAC II
CSE 486/586 Distributed Systems Gossiping
Introduction to Wireless Sensor Networks
Data Collection and Dissemination
CS 425 / ECE 428 Distributed Systems Fall 2017 Indranil Gupta (Indy)
湖南大学-信息科学与工程学院-计算机与科学系
Data Collection and Dissemination
Replica Placement Model: We consider objects (and don’t worry whether they contain just data or code, or both) Distinguish different processes: A process.
CS 425 / ECE 428 Distributed Systems Fall 2018 Indranil Gupta (Indy)
Exploring Energy-Latency Tradeoffs for Sensor Network Broadcasts
Presentation transcript:

1 CS 525 Advanced Distributed Systems Spring 09 Indranil Gupta Lecture 7 More on Epidemics (or “Tipping Point Protocols”) February 12, 2009 (gatorlog.com) (epath.org)

2 Question… What fraction of main roads need to be randomly knocked out before source and destination are completely cut off? Destination Source

3 Critical Value? Answer = 0.5 Tipping Point! Source Destination (Comes from Percolation Theory)

4 “Tipping Point” [Malcolm Gladwell, The Tipping Point, Little Brown and Company, ISBN: ] Tipping is that (magic) moment when an idea, trend or social behavior crosses a threshold, and spreads like wildfire.

5 Epidemic Protocols A specific class of tipping point protocols Local behavior at each node – probabilistic Determines global, emergent behavior at the scale of the distributed system As one tunes up the local probabilities, the global behavior may undergo a threshold behavior (or, a phase change) Three papers: 1.Epidemic algorithms 2.Bimodal multicast 3.PBBF (sensor networks)

6 Epidemic Algorithms for Replicated Database Maintenance Alan Demers et. al. Xerox Palo Alto Research Center PODC 1987 [Some slides borrowed from presentation by: R. Ganti and P. Jayachandran]

7 Introduction Maintain mutual consistency of updates in a distributed and replicated database Used in Clearinghouse database – developed in Xerox PARC and used for many years First cut approaches –Direct mail: send updates to all nodes Timely and efficient, but unreliable –Anti-entropy: exchange database content with random site Reliable, but slower than direct mail and uses more resources –Rumor mongering: exchange only ‘hot rumor’ updates Less reliable than anti-entropy, but uses fewer resources

8 Epidemic Multicast Protocol rounds (local clock) Protocol rounds (local clock) b random targets per round b random targets per round Uninfected Uninfected Infected Infected Gossip Message (UDP) (from Lecture 1)

9 Epidemic Multicast (Push) Protocol rounds (local clock) Protocol rounds (local clock) b random targets per round b random targets per round Uninfected Uninfected Infected Infected Gossip Message (UDP)

10 Epidemic Multicast (Pull) Protocol rounds (local clock) Protocol rounds (local clock) b random targets per round b random targets per round Uninfected Uninfected Infected Infected Gossip Message (UDP)

11 Pull > Push Pull converges faster than push, thus providing better delay Push-pull hybrid variant possible (see Karp and Shenker’s “Randomized Rumor Spreading”) Pull Push p i – Probability that a node is susceptible after the i th round

12 Anti-entropy: Optimizations Maintain checksum, compare databases if checksums unequal Maintain recent update lists for time T, exchange lists first Maintain inverted index of database by timestamp; exchange information in reverse timestamp order, incrementally re-compute checksums

13 Epidemic Flavors Blind vs. Feedback –Blind: lose interest to gossip with probability 1/k every time you gossip –Feedback: Loss of interest with probability 1/k only when recipient already knows the rumor Counter vs. Coin –Coin: above variants –Counter: Lose interest completely after k unnecessary contacts. Can be combined with blind. Push vs. Pull

14 Deletion and Death Certificates Absence of item does not spread; On the contrary, it can get resurrected! Use of death certificates (DCs) – when a node receives a DC, old copy of data is deleted How long to maintain a DC? –Typically twice (or some multiple of) the time to spread the information –Alternately, use Chandy and Lamport snapshot algorithm to ensure all nodes have received –Certain sites maintain dormant DCs for a longer duration; re-awakened if item seen again

15 Performance Metrics Residue: Fraction of susceptibles left when epidemic finishes Traffic: (Total update traffic) / (No. of sites) Delay: Average time for receiving update and maximum time for receiving update Some results: –Counters and feedback improve delay –Pull provides lower delay than push

16 Performance Evaluation Tipping Point Behavior

17 Discussion Pick your favorite: Push vs. pull vs. push-pull –Name one disadvantage of each Direct mail vs. anti-entropy vs. rumor mongering –Name one disadvantage of each Random neigbhor picking –Disadvantage in wired networks? –In Sensor network?

18 Bimodal Multicast Kenneth P. Birman et. al. ACM TOCS 1999 [Some slides borrowed from presentation by: W. Fagen and L. Cook]

19 “Traditional” Multicast Protocols

20 Vs. Pbcast Atomicity: All or none delivery Multicast stability: Reliable immediately delivery of messages Scalability: Bad. Costs >= quadratic with group size. Ordering Atomicity: Bimodal delivery guarantee, almost all or almost none (immediately) Multicast stability: Reliable eventual delivery of messages Scalability: Costs logarithmic w.r.t. network size. Throughput stability. Ordering Traditional MulticastPbcast

21 Pbcast: Probabilistic Broadcast Protocol Pbcast has two stages: 1.Unreliable, hierarchical, best-effort broadcast. Eg. IP Multicast 2.Two-phase anti-entropy protocol: runs simultaneously with the broadcast messages First phase detects message loss Second phase corrects such losses

22 The second stage Anti-entropy round: –Gossip Messages: Each process chooses another random process and sends a summary of its recent messages –Solicitation Messages: Messages sent back to the sender of the gossip message requesting a resend of a given set of messages (not necessarily the original source) –Message Resend: Upon reception of a solicitation message, the sender resends that message Protocol parameters at each node –# of rounds and # of processes contacted in each round –Product of above two parameters called fanout

23 Optimizations Soft-Failure Detection: Retransmission requests served only if received recently; protects against congestion caused due to redundant retransmissions Round Retransmission Limit: Limit the no. of retransmissions in a round; spread overhead in space and time Most-Recent-First Retransmission: prefer recent messages Independent Numbering of Rounds: Allows delivery and garbage collection to be entirely a local decision Multicast for Some Retransmissions

24 Bimodality of Pbcast Almost noneAlmost all Logarithmic Y-axis

25 Latency for Delivery Logarithmic growth

26 Throughput Comparison

27 Discussion Disadvantages of Bimodal Multicast? –When would wasteful messages be sent? What happens when –Rate of injection of multicasts is very very low? –IP multicast is very very reliable? –IP multicast is very very unreliable?

28 PBBF: Probability-Based Broadcast Forwarding Cigdem Sengul and Matt Miller ICDCS 2005 and ACM TOSN 2008 (Originated from a 525 Project)

29 Broadcast in an Ad-Hoc Network Ad-hoc sensor network (Grid example below) One node has a piece of information that it needs to broadcast: e.g., (1) code update, (2) query Simple approach: each node floods received message to all its neighbors –Disadvantages?

30 IEEE PSM A real, stable MAC protocol (similar results for S- MAC, T-MAC, etc.) Nodes are assumed to be synchronized Every beacon interval (BI), all nodes wake up for an ATIM window (AW) During the AW, nodes advertise any traffic that they have queued After the AW, nodes remain active if they expect to send or receive data based on advertisements; otherwise nodes return to sleep until the next BI

31 Protocol Extreme #1 A N1 N2 N3 D A = ATIM Pkt D = Data Pkt N2N1N3 A D A

32 Protocol Extreme #2 N1 N2 N3 D A = ATIM Pkt D = Data Pkt D D A N2N1N3

33 Probability-Based Broadcast Forwarding (PBBF) Introduce two parameters to sleep scheduling protocols: p and q When a node is scheduled to sleep, it will remain active with probability q When a node receives a broadcast, it rebroadcasts immediately with probability p –With probability (1-p), the node will wait and advertise the packet during the next AW before rebroadcasting the packet

34 Analysis: Reliability Phase transition when: pq + (1-p) ≈ Larger than traditional bond percolation threshold –Boundary effects –Different metric Still shows phase transition q p=0.25 p=0.37 p=0.5 p=0.75 Fraction of Broadcasts Received by 99% of Nodes Tipping Point!

35 Application: Energy and Latency Energy Joules/Broadcast q Latency Average 5-Hop Latency PBBF Increasing p q ≈ 1 + q * [(BI - AW)/AW] Ns2 simulation: 50 nodes, uniform placement, 10 avg. neighbors

36 Adaptive PBBF Energy Latency Achievable Region

37 Adaptive PBBF (TOSN paper) Dynamically adjusting p and q to converge to user- specified QoS metrics –Code updates prefer reliability overl latency –Queries prefer latency over reliability Can specify any 2 of energy, latency, and reliability Subject to those constraints, p and q are adjusted to achieve the highest reliability possible Time q p

38 Discussion PBBF: bond percolation (remove roads from city) Haas et al paper (Infocom): site percolation –Remove intersections/junctions (not roads) from city Site percolation and bond percolation have different thresholds and behaviors Hybrid possible? (like push-pull?) What about over-hearing optimizations? (like feedback)

39 Question… Are there other tipping point protocols…? Destination Source

40 Next Week Onwards Student Presentations start (see instructions) Reviews needed (see instructions) Project Meetings start (see newsgroup) –Think about which testbed you need access to: PlanetLab, Emulab, Cirrus Tomorrow: Yahoo! Training seminar