CAESAR: Context-Aware Event Stream Analytics in Real time Olga - - PowerPoint PPT Presentation

β–Ά
caesar context aware event stream analytics in real time
SMART_READER_LITE
LIVE PREVIEW

CAESAR: Context-Aware Event Stream Analytics in Real time Olga - - PowerPoint PPT Presentation

Jul 09, 2023 211 likes β€’571 views

CAESAR: Context-Aware Event Stream Analytics in Real time Olga Poppe, Chuan Lei, Elke A. Rundensteiner, and Dan Dougherty March 18, 2016 1 Complex Event Processing CEP engine Primitive events Complex events 1 , 2 , 3 The

slide-1
SLIDE 1

CAESAR: Context-Aware Event Stream Analytics in Real time

Olga Poppe, Chuan Lei, Elke A. Rundensteiner, and Dan Dougherty

March 18, 2016

1

slide-2
SLIDE 2

Worcester Polytechnic Institute

CEP engine Complex Event Processing

The same workload of independent event queries is continuously evaluated

2

𝑅1, 𝑅2, 𝑅3

Primitive events Complex events

slide-3
SLIDE 3

Worcester Polytechnic Institute

Application Context

  • Event compositions signify application contexts
  • Most event queries are appropriate only in certain contexts
  • They can be safely suspended otherwise

Examples of application contexts:

  • Emergency management: normal, crowded, fire
  • Health care: safe, warning, violation
  • Algorithmic trading: hold, buy, sell
  • Financial fraud: approved, suspicious, fraud

3

slide-4
SLIDE 4

Worcester Polytechnic Institute

Traffic Management Use Case

4

  • 140 hours idling in traffic due to congestion in 10-worst

U.S. traffic corridors per year [The Wall Street Journal]

  • Health cost of $18 billion due to traffic noise and pollution

in the USA's 83 largest urban areas in 2010 [USA Today]

  • 1.24 million deaths due to traffic injuries worldwide in

2010 [Wikipedia]

slide-5
SLIDE 5

Worcester Polytechnic Institute

Traffic Management Contexts

5

Accident

Congestion Clear Goal is to leverage application contexts to speed up system responsiveness

Accident warning Route re-computation Toll notification Route re-computation Statistics Local services

slide-6
SLIDE 6

Worcester Polytechnic Institute

Challenges

  • Rich semantics

─ Complex conditions implying a context ─ Unknown and unbounded context duration ─ Multiple inter-dependent event queries

  • Readable specification
  • Real time responsiveness

6

slide-7
SLIDE 7

Worcester Polytechnic Institute

State-of-the-art Approaches

7

CEP Systems (Esper, StreamInsight)

CAESAR

Business Models (BPMN, UML)

Expressive event queries Application contexts Context- aware

  • ptimizations
slide-8
SLIDE 8

Worcester Polytechnic Institute

Contributions & Outline

8

CAESAR system:

  • Graphical model
  • Context-aware algebra
  • Context-driven optimization techniques
  • Execution infrastructure

Performance evaluation

slide-9
SLIDE 9

Worcester Polytechnic Institute

Outline

CAESAR Model

9

slide-10
SLIDE 10

Worcester Polytechnic Institute

Context-aware Event Stream Analytics

10

slide-11
SLIDE 11

Worcester Polytechnic Institute

11

Context-aware Event Stream Analytics

slide-12
SLIDE 12

Worcester Polytechnic Institute

12

Context-aware Event Stream Analytics

slide-13
SLIDE 13

Worcester Polytechnic Institute

13

Application Contexts

slide-14
SLIDE 14

Worcester Polytechnic Institute

14

Context Deriving Queries

slide-15
SLIDE 15

Worcester Polytechnic Institute

15

Context Processing Queries

slide-16
SLIDE 16

Worcester Polytechnic Institute

Context-aware Event Queries

16

slide-17
SLIDE 17

Worcester Polytechnic Institute

Outline

CAESAR Algebra

17

slide-18
SLIDE 18

Worcester Polytechnic Institute

Context-preserving Plan Generation

18

slide-19
SLIDE 19

Worcester Polytechnic Institute

CAESAR Algebra Operators

  • 1. Context initiation 𝐷𝐽c 𝐽, 𝑋
  • 2. Context termination π·π‘ˆ

c 𝐽, 𝑋

  • 3. Context window 𝐷𝑋

𝑑 𝐽, 𝑋

  • 4. Filter πΊπ½πœ„(𝐽)
  • 5. Projection 𝑄𝑆𝐡,𝐹(𝐽)
  • 6. Event pattern 𝑄(𝐽)

19

slide-20
SLIDE 20

Worcester Polytechnic Institute

Runtime Context Maintenance

20

Context bit vector 𝑋: Context types: Time stamp 𝑋. 𝑒𝑗𝑛𝑓

1 1

𝑑a, cb, … cz

  • Updated by the context initiation & termination operators
  • Accessed by the context window operator
  • Synchronized by the time driven scheduler
slide-21
SLIDE 21

Worcester Polytechnic Institute

21

Translation from Query Set to Algebra Plan

DERIVE Toll(c.id, c.sec, 5) PATTERN NewCar c CONTEXT congestion DERIVE NewCar(s.id, s.xway, s.dir, s.seg, s.lane, s.pos, s.lane) PATTERN SEQ(NOT Position f, Position s) WHERE f.sec+30=s.sec AND f.id=s.id AND f.lane≠′ exit′ CONTEXT congestion

slide-22
SLIDE 22

Worcester Polytechnic Institute

Outline

CAESAR Optimizer

22

slide-23
SLIDE 23

Worcester Polytechnic Institute

CAESAR Optimizer Overview

23

Problem statement:

Given a workload of context-aware event queries,

  • ur optimization problem is to find an optimized query

plan for this workload with minimal CPU cost.

Context-aware optimization techniques:

  • Context window push down strategy
  • Context workload sharing algorithm
slide-24
SLIDE 24

Worcester Polytechnic Institute

Context Window Push Down Strategy

24

Performance benefits:

  • Suspension of irrelevant operators
  • Context-driven stream routing
slide-25
SLIDE 25

Worcester Polytechnic Institute

Context Workload Sharing Algorithm

25

slide-26
SLIDE 26

Worcester Polytechnic Institute

Context Workload Sharing Algorithm

26

slide-27
SLIDE 27

Worcester Polytechnic Institute

Context Workload Sharing Algorithm

27

slide-28
SLIDE 28

Worcester Polytechnic Institute

Outline

CAESAR Infrastructure & Experiments

28

slide-29
SLIDE 29

Worcester Polytechnic Institute

CAESAR Architecture

29

slide-30
SLIDE 30

Worcester Polytechnic Institute

Experimental Setup

30

Execution infrastructure: Java 7, 1 Linux machine with 16-core 3.4 GHz CPU and 48GB of RAM Data sets:

  • Linear Road stream benchmark (LR) [1]

3 roads=1.7GB

  • Physical Activity Monitoring real data set (PAM) [2]

1.6GB

[1] A.Arasu et al., Linear Road: A stream data management benchmark. VLDB’04 [2] A.Reiss et al., Creating and benchmarking a new data set for physical activity monitoring. PETRA’12

slide-31
SLIDE 31

Worcester Polytechnic Institute

Context-aware Event Stream Analytics

31

For 7 roads, context-aware (CA) event stream analytics is 9-fold faster than context-independent (CI) approach.

slide-32
SLIDE 32

Worcester Polytechnic Institute

Context-aware Event Query Sharing

32

If 30 context windows of length 15 minutes process 4 event queries each and overlap by 15 minutes, workload sharing wins 6-fold.

slide-33
SLIDE 33

Worcester Polytechnic Institute

Outline

Conclusions

33

slide-34
SLIDE 34

Worcester Polytechnic Institute

Conclusions

34

  • CAESAR is first context-aware CEP system
  • Graphical context-specification model
  • Context-aware algebra
  • Context-driven optimization techniques
  • Execution infrastructure
  • 8-fold speed up on average
slide-35
SLIDE 35

Worcester Polytechnic Institute

Acknowledgement

35

  • Advisors: Elke A. Rundensteiner, Dan Dougherty
  • Collaborator: Chuan Lei
  • DSRG group at WPI
  • EDBT reviewers
  • NSF grants IIS 1018443 and IIS 1343620