WHAT WE TALK ABOUT WHEN WE TALK ABOUT DISTRIBUTED SYSTEMS ALVARO - - PowerPoint PPT Presentation

▶
what we talk about when we talk about distributed systems
SMART_READER_LITE
LIVE PREVIEW

WHAT WE TALK ABOUT WHEN WE TALK ABOUT DISTRIBUTED SYSTEMS ALVARO - - PowerPoint PPT Presentation

Sep 28, 2023 387 likes •1.58k views

WHAT WE TALK ABOUT WHEN WE TALK ABOUT DISTRIBUTED SYSTEMS ALVARO VIDELA DISTRIBUTED SYSTEMS FOR THE IKEA FAMILY ALVARO VIDELA HTTP://BIT.LY/DIST-SYS101 @HINTJENS DISTRIBUTED SYSTEMS A DISTRIBUTED SYSTEM IS ONE IN WHICH THE FAILURE

slide-1
SLIDE 1

WHAT WE TALK ABOUT WHEN WE TALK ABOUT DISTRIBUTED SYSTEMS

ALVARO VIDELA

slide-2
SLIDE 2

DISTRIBUTED SYSTEMS FOR THE IKEA FAMILY

ALVARO VIDELA

slide-3
SLIDE 3

HTTP://BIT.LY/DIST-SYS101

slide-4
SLIDE 4

@HINTJENS

slide-5
SLIDE 5

DISTRIBUTED SYSTEMS

slide-6
SLIDE 6

“A DISTRIBUTED SYSTEM IS ONE IN WHICH THE FAILURE OF A COMPUTER YOU DID NOT EVEN KNOW EXISTED CAN RENDER YOUR OWN COMPUTER UNUSABLE”

Leslie Lamport

slide-7
SLIDE 7
slide-8
SLIDE 8
slide-9
SLIDE 9

Google: define jargon

slide-10
SLIDE 10

DISTRIBUTED SYSTEMS

slide-11
SLIDE 11

DISTRIBUTED SYSTEMS

  • Many entities trying to solve a problem

(nodes, processes)

slide-12
SLIDE 12

DISTRIBUTED SYSTEMS

  • Many entities trying to solve a problem

(nodes, processes)

  • Partial Knowledge
slide-13
SLIDE 13

DISTRIBUTED SYSTEMS

  • Many entities trying to solve a problem

(nodes, processes)

  • Partial Knowledge
  • Uncertainty
slide-14
SLIDE 14

DEEP RABBIT HOLE

slide-15
SLIDE 15

WHAT TO READ?

slide-16
SLIDE 16

WHICH PAPERS?

slide-17
SLIDE 17
slide-18
SLIDE 18
slide-19
SLIDE 19
slide-20
SLIDE 20
slide-21
SLIDE 21
slide-22
SLIDE 22

WHICH BOOKS?

slide-23
SLIDE 23
slide-24
SLIDE 24

WHY?

slide-25
SLIDE 25

http://tobielangel.com

slide-26
SLIDE 26

THE PROBLEM

slide-27
SLIDE 27

DIFFERENT MODELS

slide-28
SLIDE 28

DIFFERENT MODELS

  • Timing Model
slide-29
SLIDE 29

DIFFERENT MODELS

  • Timing Model
  • Inter Process Communication Used (IPC

method)

slide-30
SLIDE 30

DIFFERENT MODELS

  • Timing Model
  • Inter Process Communication Used (IPC

method)

  • Failure Modes
slide-31
SLIDE 31

TIMING MODEL

slide-32
SLIDE 32

TIMING MODEL

  • Synchronous Model
slide-33
SLIDE 33

TIMING MODEL

  • Synchronous Model
  • Asynchronous Model
slide-34
SLIDE 34

TIMING MODEL

  • Synchronous Model
  • Asynchronous Model
  • Semi-synchronous Model
slide-35
SLIDE 35

INTERPROCESS COMMUNICATION

slide-36
SLIDE 36

INTERPROCESS COMMUNICATION

  • Message Passing
slide-37
SLIDE 37

INTERPROCESS COMMUNICATION

  • Message Passing
  • Shared Memory
slide-38
SLIDE 38

FAILURE MODES

slide-39
SLIDE 39

FAILURE MODES

  • Crash-stop
slide-40
SLIDE 40

FAILURE MODES

  • Crash-stop
  • Crash-recovery
slide-41
SLIDE 41

FAILURE MODES

  • Crash-stop
  • Crash-recovery
  • Omission Faults
slide-42
SLIDE 42

FAILURE MODES

  • Crash-stop
  • Crash-recovery
  • Omission Faults
  • Arbitrary Failures Mode (Byzantine)
slide-43
SLIDE 43

LIVENESS AND SAFETY

slide-44
SLIDE 44

LIVENESS AND SAFETY PROPERTIES OF ALGORITHMS

slide-45
SLIDE 45

SAFETY

Some “bad” thing does not happens during execution

slide-46
SLIDE 46

SAFETY

“Communication links should not invent messages out of thin air”

slide-47
SLIDE 47

LIVENESS

A “good” thing happens during execution

slide-48
SLIDE 48

LIVENESS

“A destination process eventually delivers the message”

slide-49
SLIDE 49

LET’S TAKE A LOOK AT FLP

1

1 - Fischer, Lynch, Paterson

slide-50
SLIDE 50
slide-51
SLIDE 51

IMPOSSIBILITY OF DISTRIBUTED CONSENSUS WITH ONE FAULTY PROCESS

slide-52
SLIDE 52

IMPOSSIBILITY OF DISTRIBUTED CONSENSUS WITH ONE FAULTY PROCESS

slide-53
SLIDE 53

IMPOSSIBILITY OF DISTRIBUTED CONSENSUS WITH ONE FAULTY PROCESS

slide-54
SLIDE 54

IMPOSSIBILITY OF DISTRIBUTED CONSENSUS WITH ONE FAULTY PROCESS

slide-55
SLIDE 55

IMPOSSIBILITY OF DISTRIBUTED CONSENSUS WITH ONE FAULTY PROCESS

slide-56
SLIDE 56

WHAT’S CONSENSUS ANYWAY?

slide-57
SLIDE 57

“THE CONSENSUS PROBLEM IS A PARADIGM OF AGREEMENT PROBLEMS”

https://dl.acm.org/citation.cfm?id=1052796.1052806

slide-58
SLIDE 58

PROPERTIES OF CONSENSUS

slide-59
SLIDE 59

PROPERTIES OF CONSENSUS

  • C-Termination: Every correct process eventually decides on some value
slide-60
SLIDE 60

PROPERTIES OF CONSENSUS

  • C-Termination: Every correct process eventually decides on some value
  • C-Validity: If a process decides v, then v was proposed by some process
slide-61
SLIDE 61

PROPERTIES OF CONSENSUS

  • C-Termination: Every correct process eventually decides on some value
  • C-Validity: If a process decides v, then v was proposed by some process
  • C-Agreement: No two correct processes decide differently
slide-62
SLIDE 62

PROPERTIES OF UNIFORM CONSENSUS

  • C-Termination: Every correct process eventually decides on some value
  • C-Validity: If a process decides v, then v was proposed by some process
  • C-Agreement: No two correct processes decide differently
  • C-Uniform Agreement: No two processes (correct or not) decide

differently.

slide-63
SLIDE 63

WE NEED CONSENSUS WHEN:

A SET OF PROCESSES HAVE TO AGREE TO TAKE A COMMON ACTION

slide-64
SLIDE 64

WE NEED CONSENSUS WHEN:

A SET OF PROCESSES HAVE TO AGREE TO TAKE A COMMON ACTION

Atomic Broadcast

slide-65
SLIDE 65

WE NEED CONSENSUS WHEN:

A SET OF PROCESSES HAVE TO AGREE TO TAKE A COMMON ACTION

Atomic Broadcast Group Membership

slide-66
SLIDE 66

ATOMIC BROADCAST

“CORRECT PROCESSES DELIVER THE SAME SET OF MESSAGES IN THE SAME ORDER”

slide-67
SLIDE 67

FLP TELLS US THAT IF CONSENSUS CANNOT BE ACHIEVED, THEN ATOMIC BROADCAST OR GROUP MEMBERSHIP CANNOT BE ACHIEVED EITHER

slide-68
SLIDE 68

SO, WE PACK OUR BAGS AND GO? NOTHING TO SEE HERE?

slide-69
SLIDE 69

STUMBLING OVER CONSENSUS RESEARCH: MISUNDERSTANDING AND ISSUES

Marcos K. Aguilera

slide-70
SLIDE 70

FAILURE DETECTORS

slide-71
SLIDE 71
slide-72
SLIDE 72

FAILURE DETECTORS

slide-73
SLIDE 73

FAILURE DETECTORS

  • External process
slide-74
SLIDE 74

FAILURE DETECTORS

  • External process
  • Provides information about suspected processes
slide-75
SLIDE 75

FAILURE DETECTORS

  • External process
  • Provides information about suspected processes
  • Completeness property (crashed processes are

detected)

slide-76
SLIDE 76

FAILURE DETECTORS

  • External process
  • Provides information about suspected processes
  • Completeness property (crashed processes are

detected)

  • Accuracy (correct process are never suspected)
slide-77
SLIDE 77

“RUB SOME PERFECT FAILURE DETECTOR ON IT”

slide-78
SLIDE 78

http://www.amazon.com/Introduction-Reliable-Secure- Distributed-Programming/dp/3642152597

PERFECT FAILURE DETECTOR

slide-79
SLIDE 79

EVENTUALLY ACCURATE FAILURE DETECTOR

slide-80
SLIDE 80

EVENTUALLY ACCURATE FAILURE DETECTOR

  • Strong Completeness: Eventually, every

process that crashes is permanently suspected by every correct process.

slide-81
SLIDE 81

EVENTUALLY ACCURATE FAILURE DETECTOR

  • Strong Completeness: Eventually, every

process that crashes is permanently suspected by every correct process.

  • Eventual Weak Accuracy: There is a time

after which some correct process is never suspected by the correct processes.

slide-82
SLIDE 82

EVENTUALLY ACCURATE FAILURE DETECTOR

  • Strong Completeness: Eventually, every

process that crashes is permanently suspected by every correct process.

  • Eventual Weak Accuracy: There is a time

after which some correct process is never suspected by the correct processes. http://dl.acm.org/citation.cfm?id=1052806

slide-83
SLIDE 83
slide-84
SLIDE 84

QUORUMS

slide-85
SLIDE 85

TL;DR:

INTERSECTING SETS

slide-86
SLIDE 86

“A QUORUM IN A SYSTEM WITH N CRASH-FAULT PROCESS ABSTRACTIONS […] IS ANY MAJORITY OF PROCESSES, I.E., ANY SET OF MORE THAN N/2 PROCESSES”

QUORUMS

slide-87
SLIDE 87

“IF F < N/2 PROCESSES FAIL BY CRASHING, THERE IS ALWAYS AT LEAST ONE QUORUM OF NONCRASHED PROCESSES IN SUCH SYSTEMS”

QUORUMS

slide-88
SLIDE 88

QUORUMS

A - B - C - D - E

slide-89
SLIDE 89

QUORUMS

A - B - C - D - E

slide-90
SLIDE 90

QUORUMS

A - B’ - C - D’ - E

slide-91
SLIDE 91

QUORUMS

A - B’ - C - D’ - E

slide-92
SLIDE 92

QUORUMS

A - B’ - C’ - D’ - E’

slide-93
SLIDE 93

QUORUMS

A - B’ - C’ - D’ - E’

slide-94
SLIDE 94

CONSISTENCY

slide-95
SLIDE 95
slide-96
SLIDE 96

CONCURRENT FIFO QUEUE

slide-97
SLIDE 97

CONSISTENCY CONDITIONS

slide-98
SLIDE 98

CONSISTENCY CONDITIONS

  • Atomic Consistency (Linearizabilty)
slide-99
SLIDE 99

CONSISTENCY CONDITIONS

  • Atomic Consistency (Linearizabilty)
  • Sequential Consistency
slide-100
SLIDE 100

CONSISTENCY CONDITIONS

  • Atomic Consistency (Linearizabilty)
  • Sequential Consistency
  • Causal Consistency
slide-101
SLIDE 101

CONSISTENCY CONDITIONS

  • Atomic Consistency (Linearizabilty)
  • Sequential Consistency
  • Causal Consistency

https://aphyr.com/posts/313-strong-consistency- models

slide-102
SLIDE 102

LINEARIZABILTY

http://www.amazon.com/Distributed-Algorithms-Message-Passing-Systems-Michel/dp/ 3642381227/

slide-103
SLIDE 103

LINEARIZABILTY

http://www.amazon.com/Distributed-Algorithms-Message-Passing-Systems-Michel/dp/ 3642381227/

slide-104
SLIDE 104

SOME BOOKS

slide-105
SLIDE 105

http://www.amazon.com/Distributed-Algorithms- Message-Passing-Systems-Michel/dp/3642381227/

slide-106
SLIDE 106

http://www.amazon.com/Fault-tolerant-Agreement- Synchronous-Message-passing-Distributed/dp/ 1608455254/

slide-107
SLIDE 107

http://www.amazon.com/Communication-Abstractions- Fault-tolerant-Asynchronous-Distributed/dp/160845293X/

slide-108
SLIDE 108

http://www.amazon.com/Distributed-Algorithms- Kaufmann-Management-Systems/dp/1558603484/

slide-109
SLIDE 109

http://www.amazon.com/Introduction-Reliable-Secure- Distributed-Programming/dp/3642152597

slide-110
SLIDE 110

http://www.amazon.com/Guide-Reliable-Distributed- Systems-High-Assurance/dp/1447124154/

slide-111
SLIDE 111

http://www.amazon.com/Replication-Practice-Lecture- Computer-Theoretical/dp/3642112935/

slide-112
SLIDE 112

FINDING NON PAYWALLED PAPERS

slide-113
SLIDE 113

CONCLUSION

slide-114
SLIDE 114

CONCLUSION

  • Deep Rabbit Hole
slide-115
SLIDE 115

CONCLUSION

  • Deep Rabbit Hole
  • Computing Science where Science is Still a Thing™
slide-116
SLIDE 116

CONCLUSION

  • Deep Rabbit Hole
  • Computing Science where Science is Still a Thing™
  • History of the Field Matters
slide-117
SLIDE 117

CONCLUSION

  • Deep Rabbit Hole
  • Computing Science where Science is Still a Thing™
  • History of the Field Matters
  • Read, read, read
slide-118
SLIDE 118

THANKS!

@old_sound