550 likes | 737 Views
Multicast Data Dissemination. Wang Lam Special University Oral Examination 7 July 2004. Contents. Current multicast networks Contributions Data scheduling Network issues Related and future work Conclusion. Traditional data service: one-to-one Multicast networks: one-to-many
E N D
Multicast Data Dissemination Wang Lam Special University Oral Examination 7 July 2004
Contents • Current multicast networks • Contributions • Data scheduling • Network issues • Related and future work • Conclusion
Traditional data service: one-to-one Multicast networks: one-to-many IP: multicast group addresses (IPv4: 224.0.0.0 - 239.255.255.255; IPv6: FF00::/8) Network bottlenecks Client joins Unreliable delivery Datagrams (UDP) Current multicast networks
Client joins Unreliable delivery Datagrams Supports varying bandwidth clients All requested data must arrive Data arranged to optimize performance Multicast data dissemination
Principal contributions • Data scheduling • Minimize delay for clients requesting many items • Scheduling for subscribers and downloaders 1 • Networking issues • Reliable delivery 2 • Splitting bandwidth into channels
Principal contributions • Data scheduling • Scheduling for subscribers and downloaders
Subscribers and downloaders • Data scheduling • Scheduling for subscribers and downloaders • Distributing data for a Web repository • Metrics and techniques • Sample results
The multicast source • Stanford WebBase • 100+ million Web pages • Additional benefits of multicast WWW multicast server clients crawler repository indexing and analysis
A multicast facility • Clients issue requests to server • Clients listen to shared multicast • Server schedules data onto multicast • Downloaders and subscribers Multicast server clients
Clients request multiple items • Broadcast disks: one-item “response time” • Multicast: client delay is different • Subscribers: freshness and age
Example scheduler: Circ • Arbitrarily order data items • Send requested data
Example scheduler: Pop • Send most requested data
Example scheduler: R/Q • Number of requesting clients • Smallest request size
Example scheduler: R/Q • Number of requesting clients • Smallest request size
Some results for subscribers • Choice of scheduler depends on performance metric • Update frequency has little effect
Summary • Differences from broadcast disks • Downloaders and subscribers • Studied design tradeoffs for various metrics and techniques
Principal contributions • Data scheduling • Minimize delay for clients requesting many items • Scheduling for subscribers and downloaders • Networking issues • Reliable delivery 2 • Splitting bandwidth into channels
Principal contributions • Networking issues • Reliable delivery
Principal contributions • Networking issues • Reliable delivery • Multicast server model • Reliability techniques • Sample results • Other challenges
The multicast source • Stanford WebBase • 100+ million Web pages • Network loss <5% to >20% WWW multicast server clients crawler repository indexing and analysis
A multicast facility • Clients issue requests to server • Clients listen to shared multicast • Server schedules data onto multicast • Data channel unreliable multicast server clients
Forward Error Correction • Compute fixed fraction of redundant data • Reconstruct from subset of bits • Vary padding by item data FEC data FEC requests requests
Forward Error Correction • Compute fixed fraction of redundant data • Reconstruct from subset of bits • Vary padding by item
Forward Error Correction • Compute fixed fraction of redundant data • Reconstruct from subset of bits • Vary padding by item
Forward Error Correction • Compute fixed fraction of redundant data • Reconstruct from subset of bits • Vary padding by item
Forward Error Correction • Compute fixed fraction of redundant data • Reconstruct from subset of bits • Vary padding by item FEC(0.2R)
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs data data requests requests
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs NAK NAK
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs
Retransmission • Wait for NAK • Queue retransmission of enough bits • Queue only on selected NAKs NAK NAK R(1)
Rescheduling • Do nothing • Rerequest data item • Combine with prior reliability schemes data data requests requests
Rescheduling • Do nothing • Rerequest data item • Combine with prior reliability schemes NAK NAK
Additional results • Error-correcting packets help retransmissions • Variable FEC can outperform matched-rate FEC • Data-in-progress announcement can slightly help new clients
Summary • Multicast server scenario allows a variety of reliability techniques • Techniques form many combinations • Studied design tradeoffs
Principal contributions • Data scheduling • Minimize delay for clients requesting many items • Scheduling for subscribers and downloaders • Networking issues • Reliable delivery • Splitting bandwidth into channels
Publications • W. Lam and H. Garcia-Molina, “Multicasting a Data Repository,” WebDB 2001 • W. Lam and H. Garcia-Molina, “Multicasting a Changing Repository,” ICDE 2003 • W. Lam and H. Garcia-Molina, “Reliably Networking a Multicast Repository,” SRDS 2003 • W. Lam and H. Garcia-Molina, “Slicing Broadcast Disks,” submitted for publication • W. Lam and H. Garcia-Molina, “Implementing Multicast Data Dissemination,” technical report
Publications (Stanford WebBase) • J. Cho, T. Haveliwala, W. Lam, S. Raghavan, A. Paepcke, and H. Garcia-Molina, “Stanford WebBase Components and Applications”http://www-diglib.stanford.edu/~testbed/doc2/WebBase/Web crawler and client code:ftp://db.stanford.edu/pub/digital_library/
Related work • Broadcast disks • Web caching • Publish/subscribe systems • Video on demand • Reliable multicast protocols • Layered multicast protocols
Next steps • Other kinds of clients • On-the-fly processing • Partially ordered clients • Opportunistic clients • Distributed servers • Request mining