prior work
play

Prior Work Consensus Consensus Reliable BGP Consensus Reliable - PowerPoint PPT Presentation

Prior Work Consensus Consensus Reliable BGP Consensus Reliable BGP Consensus Routing Reliable BGP Routing Reliable BGP Routing Routing Graceful Seamless Graceful Seamless Seamless Graceful Graceful Seamless Migration Migration


  1. Prior Work Consensus Consensus Reliable BGP Consensus Reliable BGP Consensus Routing Reliable BGP Routing Reliable BGP Routing Routing Graceful Seamless Graceful Seamless Seamless Graceful Graceful Seamless Migration Migration Migration Migration Migration Migration Migration Migration 4

  2. Prior Work Consensus Consensus Reliable BGP Consensus Reliable BGP Consensus Routing Reliable BGP Routing Reliable BGP Routing Routing Graceful Seamless Graceful Seamless Seamless Graceful Graceful Seamless Migration Migration Migration Migration Migration Migration Migration Migration 4

  3. Universal Property Preservation Theorem: Per-packet consistent updates preserve all trace properties. Trace Property Any property of a single packet’s path through the network. Examples of Trace Properties: Loop freedom, access control, waypointing ... Trace Property Verifcation Tools: Anteater , Header Space Analysis, ConfjgChecker ... 11

  4. MECHANISMS 13

  5. Correctness Question : How do we convince ourselves these mechanisms are correct? Solution : We built an operational semantics, formalized our mechanisms and proved them correct Example: 2-Phase Update } 1. Install new rules on internal switches, leave Unobservable old confjguration in place 2. Install edge rules that stamp with the new } One-touch version number 18

  6. Correctness Question : How do we convince ourselves these mechanisms are correct? Solution : We built an operational semantics, formalized our mechanisms and proved them correct Example: 2-Phase Update } 1. Install new rules on internal switches, leave Unobservable old confjguration in place 2. Install edge rules that stamp with the new } One-touch version number Theorem: Unobservable + one-touch = per-packet. 18

  7. IMPLEMENTATION & EVALUATION 19

  8. Evaluation Question: How much extra rule space is required? Setup Topologies • Mininet VM Applications • Routing and Multicast Fattree Scenarios • Adding/removing hosts • Adding/removing links • Both at the same time Small-world Waxman 21

  9. Results: Routing Application Full Subset Fattree Small-world Waxman 22

  10. Results: Routing Application 100% Full Subset Worst-Case Rule Overhead 75% 50% 25% 0% Host Link Both Host Link Both Host Link Both Fattree Small-world Waxman 22

  11. Results: Routing Application 100% Full Subset Worst-Case Rule Overhead 75% 50% 25% 0% Host Link Both Host Link Both Host Link Both Fattree Small-world Waxman 22

  12. WRAP UP 23

  13. Conclusion Update abstractions • Per-packet • Per-fow Mechanisms • 2-Phase Update • Optimizations Implementation • Runtime • Verifjer Formal model • Network operational semantics • Universal property preservation 24

  14. Thank You! Collaborators Shrutarshi Basu (Cornell) Arjun Guha (Cornell) Stephen Gutz (Cornell) Rob Harrison (West Point) Nanxi Kang (Princeton) Naga Praveen Katta (Princeton) Chris Monsanto (Princeton) http://frenetic-lang.org Josh Reich (Princeton) Cole Schlesinger (Princeton) Robert Soulé (Cornell) Alec Story (Cornell) Nate Foster (Cornell) Mike Freedman (Princeton) Jen Rexford (Princeton) Emin Gün Sirer (Cornell) Dave Walker (Princeton) 25

  15. BACKUP SLID ES 26

Recommend


More recommend