byzantine techniques
play

Byzantine Techniques Michael George November 29, 2005 Michael - PowerPoint PPT Presentation

Overview The Byzantine Generals Problem Practical Byzantine Fault Tolerance Conclusion Byzantine Techniques Michael George November 29, 2005 Michael George Byzantine Techniques Overview The Byzantine Generals Problem Practical


  1. Overview The Byzantine General’s Problem Practical Byzantine Fault Tolerance Conclusion Byzantine Techniques Michael George November 29, 2005 Michael George Byzantine Techniques

  2. Overview The Byzantine General’s Problem Practical Byzantine Fault Tolerance Conclusion Reliability and Failure “There can be no unity without agreement, and there can be no agreement without conciliation” — Ren´ e Maowad Michael George Byzantine Techniques

  3. Overview The Byzantine General’s Problem Practical Byzantine Fault Tolerance Conclusion Reliability and Failure “There can be no unity without agreement, and there can be no agreement without conciliation” — Ren´ e Maowad We want reliable systems Until now, we’ve assumed that failures are fail-stop What happens if failures are arbitrary? Michael George Byzantine Techniques

  4. Overview The Byzantine General’s Problem Practical Byzantine Fault Tolerance Conclusion Reliability and Failure “There can be no unity without agreement, and there can be no agreement without conciliation” — Ren´ e Maowad We want reliable systems Until now, we’ve assumed that failures are fail-stop What happens if failures are arbitrary? . . . or even malicious? Michael George Byzantine Techniques

  5. Overview The Byzantine General’s Problem Practical Byzantine Fault Tolerance Conclusion Today’s Presentation We will discuss two papers that address this worst-case scenario: 1 The Byzantine General’s Problem [Lamport et. al. 1982] Phrases the problem in terms of Byzantine Generals Shows a tight upper bound on fault tolerance Explores bounds under modified assumptions Michael George Byzantine Techniques

  6. Overview The Byzantine General’s Problem Practical Byzantine Fault Tolerance Conclusion Today’s Presentation We will discuss two papers that address this worst-case scenario: 1 The Byzantine General’s Problem [Lamport et. al. 1982] Phrases the problem in terms of Byzantine Generals Shows a tight upper bound on fault tolerance Explores bounds under modified assumptions 2 Practical Byzantine Fault Tolerance [Castro and Liskov 1999] Implements fault-tolerant state-machine replication Aggressively optimizes the implementation Layers replicated NFS over state-machine Shows performance penalty is reasonable Michael George Byzantine Techniques

  7. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions The Basic Problem A group of Byzantine Generals are surrounding an enemy city. They need to jointly decide whether to attack or retreat. But some of them might be traitors. Want them to agree on a decision. Michael George Byzantine Techniques

  8. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions The Basic Problem A group of Byzantine Generals are surrounding an enemy city. They need to jointly decide whether to attack or retreat. But some of them might be traitors. Want them to agree on a decision. Decision must be good. Michael George Byzantine Techniques

  9. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Reducing Decision Making to Information Propogation If a single commander can send information to some lieutenants such that: IC1 All loyal lieutenants recieve the same value IC2 If commander is loyal, then all lieutenants recieve value she sent Michael George Byzantine Techniques

  10. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Reducing Decision Making to Information Propogation If a single commander can send information to some lieutenants such that: IC1 All loyal lieutenants recieve the same value IC2 If commander is loyal, then all lieutenants recieve value she sent Alice A Bob R Cathy A Don A Michael George Byzantine Techniques

  11. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Reducing Decision Making to Information Propogation If a single commander can send information to some lieutenants such that: IC1 All loyal lieutenants recieve the same value IC2 If commander is loyal, then all lieutenants recieve value she sent Commander Alice A Bob R Cathy A Don A A A A A Lieutenants Michael George Byzantine Techniques

  12. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Reducing Decision Making to Information Propogation If a single commander can send information to some lieutenants such that: IC1 All loyal lieutenants recieve the same value IC2 If commander is loyal, then all lieutenants recieve value she sent Commander Alice A Bob R Cathy A Don A A R A R A R A R Lieutenants Michael George Byzantine Techniques

  13. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Reducing Decision Making to Information Propogation If a single commander can send information to some lieutenants such that: IC1 All loyal lieutenants recieve the same value IC2 If commander is loyal, then all lieutenants recieve value she sent Commander Alice A Bob R Cathy A Don A A R A A R A A R A A R A Lieutenants Michael George Byzantine Techniques

  14. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Reducing Decision Making to Information Propogation If a single commander can send information to some lieutenants such that: IC1 All loyal lieutenants recieve the same value IC2 If commander is loyal, then all lieutenants recieve value she sent Commander Alice A Bob R Cathy A Don A A R A A A R A A A R A A A R A A Lieutenants Michael George Byzantine Techniques

  15. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Reducing Decision Making to Information Propogation If a single commander can send information to some lieutenants such that: IC1 All loyal lieutenants recieve the same value IC2 If commander is loyal, then all lieutenants recieve value she sent Alice A Bob R Cathy A Don A A R A A A R A A A R A A A R A A The Byzantine General’s Problem is to send information in a way that satisfies IC1 and IC2. Michael George Byzantine Techniques

  16. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Impossibility With Three Generals Consider the following: Commander “attack” “attack” he said “attack” Lieutenant 1 Lieutenant 2 Commander says Lieutenant 2 says Lieutenant 1 concludes “attack” “attack” “attack” Michael George Byzantine Techniques

  17. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Impossibility With Three Generals Consider the following: Commander “attack” “attack” he said “retreat” Lieutenant 1 Lieutenant 2 Commander says Lieutenant 2 says Lieutenant 1 concludes “attack” “attack” “attack” “attack” “retreat” “attack” Michael George Byzantine Techniques

  18. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Impossibility With Three Generals Consider the following: Commander “retreat” “attack” he said “retreat” Lieutenant 1 Lieutenant 2 Commander says Lieutenant 2 says Lieutenant 1 concludes “attack” “attack” “attack” “attack” “retreat” “attack” “attack” “retreat” “retreat” Michael George Byzantine Techniques

  19. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Impossibility With 3 m Generals What if we can solve for 3 m Albanian generals with m failures? AC AL 1 AL 2 AL 3 AL 4 AL 5 AL 6 AL 7 AL 8 Michael George Byzantine Techniques

  20. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Impossibility With 3 m Generals What if we can solve for 3 m Albanian generals with m failures? Byzantine Commander AC Byzantine Lieut. 1 Byzantine Lieut. 2 AL 1 AL 2 AL 3 AL 4 AL 5 AL 6 AL 7 AL 8 Then we can implement three Byzantine generals with one failure! Michael George Byzantine Techniques

  21. Overview The Basic Problem The Byzantine General’s Problem Impossibility Results Practical Byzantine Fault Tolerance An Optimal Algorithm Conclusion Extensions Impossibility With Approximate Agreement Can we do approximate (within a given δ ) agreement? Michael George Byzantine Techniques

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend