Big Data and Internet Thinking Chentao Wu Associate Professor - - PowerPoint PPT Presentation

▶
big data and internet thinking
SMART_READER_LITE
LIVE PREVIEW

Big Data and Internet Thinking Chentao Wu Associate Professor - - PowerPoint PPT Presentation

Jan 12, 2023 286 likes •1.12k views

Big Data and Internet Thinking Chentao Wu Associate Professor Dept. of Computer Science and Engineering wuct@cs.sjtu.edu.cn Download lectures ftp://public.sjtu.edu.cn User: wuct Password: wuct123456

slide-1
SLIDE 1

Big Data and Internet Thinking

Chentao Wu Associate Professor

  • Dept. of Computer Science and Engineering

wuct@cs.sjtu.edu.cn

slide-2
SLIDE 2

Download lectures

  • ftp://public.sjtu.edu.cn
  • User: wuct
  • Password: wuct123456
  • http://www.cs.sjtu.edu.cn/~wuct/bdit/
slide-3
SLIDE 3

Schedule

  • lec1: Introduction on big data, cloud computing & IoT
  • Iec2: Parallel processing framework (e.g., MapReduce)
  • lec3: Advanced parallel processing techniques (e.g.,

YARN, Spark)

  • lec4: Cloud & Fog/Edge Computing
  • lec5: Data reliability & data consistency
  • lec6: Distributed file system & objected-based storage
  • lec7: Metadata management & NoSQL Database
  • lec8: Big Data Analytics
slide-4
SLIDE 4

Collaborators

slide-5
SLIDE 5

Contents

  • Intro. to Cloud/Fog Computing

1

slide-6
SLIDE 6

Fog Computing

slide-7
SLIDE 7

Challenges

slide-8
SLIDE 8

Cloud-Fog-Edge

slide-9
SLIDE 9

Cloud-Fog-Edge Devices

slide-10
SLIDE 10

Cloud-Fog-Edge Architecture (1)

slide-11
SLIDE 11

Cloud-Fog-Edge Architecture (2)

slide-12
SLIDE 12

Functions of Cloud-Fog-Edge

slide-13
SLIDE 13

Fog/Edge Computing is the Primary Choice to Handle Real Time Data

slide-14
SLIDE 14

IoT End-to-End Value Chain

slide-15
SLIDE 15

IoT in the cloud and on the edge

IoT in the Cloud

▪ Remote monitoring and control ▪ Merging remote data from across multiple IoT devices ▪ Near infinite compute and storage to train machine learning and other advanced AI tools

IoT on the Edge

➔

Low latency tight control loops require near real-time response

➔

Public internet inherently unpredictable

➔

Privacy of data and protection

  • f IP
slide-16
SLIDE 16

Data Processing in Cloud-Fog-Edge

slide-17
SLIDE 17

Heterogeneous/Homogeneous Cloud/Fog computing

slide-18
SLIDE 18

Heterogeneous/Homogeneous Computing Framework

  • Cloud: Parallel and Distributed Computing

 Map-Reduce, Graph Computing, Stream Computing

  • Edge/Fog: Approximate Computing
slide-19
SLIDE 19

Why Approximate Computing?

▪ Inexact/imprecise input data ▪ Approximate/iterative algorithms ▪ Loose constraints on output

image, sound and video processing image rendering sensor data analysis, computer vision

✓ ✓

simulations, games, search, machine learning

✓ ✓

Where a lot of (most?) resources go!

slide-20
SLIDE 20

Point of View on Approximate Computing

Performance Resource usage (e.g., energy) 2

slide-21
SLIDE 21

Approximate Computing Example - Images(1)

slide-22
SLIDE 22

Approximate Computing Example - Images(2)

Energy Errors Energy Errors Energy Errors Energy Errors

slide-23
SLIDE 23

Approximate Computing in Different Areas

Architecture OS/DB Runtime Compiler PL Hardware EnerJ (UW), Passert (MSR/UW), Rely (MIT), Relax (Wisconsin) Uncertain<T> (MSR), Eon (UMass) Probabilistic CMOS (Rice), approximate components (Purdue) ANNs (UW, MSR, INRIA, Wisconsin, Qualcomm) Stochastic Processors (UIUC), ERSA (Stanford), Flikker (MSR), QUORA (Purdue), Approximate Storage (MSR, UW) Using Neural Nets for code approximation (GAtech/UW/MSR) Stream Processing (Princeton) Green (MSR), PowerDial (MIT), soft error control (UCLA), SAGE & Paraprox (Michigan), Swat (UIUC) BlinkDB (Berkeley/MIT) Probabilistic transformations (MIT)

slide-24
SLIDE 24

Approximate Computing Using Neutral Networks

Neural Representation Code1 Code2 Code3 Code4 Code5 Code6

…

CPU NPU

Source Code Common Intermediate Representation Acceleration

+ ×

slide-25
SLIDE 25

Approximate Computing – Program (1)

Program

slide-26
SLIDE 26

Approximate Computing – Program (2)

Program

Fin ind an approximate program component

slide-27
SLIDE 27

Approximate Computing – Program (3)

Program

Compil ile the program and train a neural network Fin ind an approximate program component

slide-28
SLIDE 28

Approximate Computing – Program (4)

Program

Compil ile the program and train a neural network Execut ute on a fast Neural Processing Unit (NPU) Fin ind an approximate program component

slide-29
SLIDE 29

Approximate Computing – Neutral Network Acceleration

CPU NPU

CPU GPU FPGA Digital ASIC FPAA Analog ASIC

(Speed: ~4×↑, Energy: ~10×↓, Quality: 5%↓)

slide-30
SLIDE 30

Approximate Codes

Disciplined Approximate Programming (EnerJ, EnerC,...)

int p = 5; @Approx int a = 7; for (int x = 0..) { a += func(2); @Approx int z; z = p * 2; p += 4; } a /= 9; p += 10; socket.send(z); write(file, z);

Relaxed Algorithms

λ

Aggressive Compilation

ɸ

Approximate Data Storage Variable-Accuracy ISA

ALU

Approximate Logic/Circuits

Variable-quality wireless communication

slide-31
SLIDE 31

Contents

  • Intro. to Fog/Edge Networking

2

slide-32
SLIDE 32

Edge Architecture

slide-33
SLIDE 33

Edge-Fog-Cloud Network

slide-34
SLIDE 34

Layered Network (1)

slide-35
SLIDE 35

Layered Network (2)

slide-36
SLIDE 36

Different Requirements on Latency

slide-37
SLIDE 37

Different Network Topology

slide-38
SLIDE 38

Different Network Protocols

slide-39
SLIDE 39

Different Network Accesses

  • RAN - Radio Area

ea Netw twork for LTE/5G

  • RNC- Radio

io Netw twork Con Controll ller for

  • r WiF

iFi

  • CMTS- Ca

Cable le Mod

  • dem

Ter ermination System

  • PON OLT for fiber
  • EPC – evol
  • lved

ed Pack cketCore

slide-40
SLIDE 40

Different Network Connections

slide-41
SLIDE 41

Applications – Device Location

slide-42
SLIDE 42

Applications – Video Analytics

slide-43
SLIDE 43

Applications – Content Optimization

slide-44
SLIDE 44

Applications – DNS Caching

slide-45
SLIDE 45

Applications – Application Optimization

slide-46
SLIDE 46

Edge-Fog-Cloud Network Example: TelcoFog (1)

slide-47
SLIDE 47

Edge-Fog-Cloud Network Example: TelcoFog (2)

slide-48
SLIDE 48

Edge-Fog-Cloud Network Example: TelcoFog (3)

slide-49
SLIDE 49

Edge-Fog-Cloud Network Example: TelcoFog (4)

slide-50
SLIDE 50

Contents

Industrial Solutions

3

slide-51
SLIDE 51

Google IoT Solution

slide-52
SLIDE 52

Google Edge Computing (1)

slide-53
SLIDE 53

Google Edge Computing (2)

slide-54
SLIDE 54

Cloud & Edge Fusion – System Architecture

slide-55
SLIDE 55

Cloud & Edge Fusion – Model Training

▪ Fog: Collecting Data ▪ Cloud: Rendering & Training Data

slide-56
SLIDE 56

Cloud & Edge Fusion – Virtualization (1)

slide-57
SLIDE 57

Cloud & Edge Fusion – Virtualization (2)

slide-58
SLIDE 58

Cloud & Edge Fusion – Virtualization (3)

slide-59
SLIDE 59

Cloud & Edge Fusion – Virtualization (4)

slide-60
SLIDE 60

Amazon AWS IoT Solution

▪ FreeRTOS: IoT operating system ▪ Greengrass： Seamless expansion to edge devices

slide-61
SLIDE 61

Amazon AWS IoT Architecture

AWS IoT Architecture

Endpoints

Fleet onboarding, management and SW updates Fleet audit and protection IoT data analytics and intelligence

Gateway

Things Sense & Act Cloud Storage & Compute

Secure local triggers, actions, and data sync

Intelligence Insights & Logic → Action

Secure device connectivity and messaging

slide-62
SLIDE 62

Free RTOS - an open source IoT OS

▪ FreeRTOS: https://www.freertos.org/

slide-63
SLIDE 63

Greengrass – AWS Edge Computing Platform (1)

▪ Greengrass is an edge/ fog node with certain computing and processing capability in AWS

slide-64
SLIDE 64

Greengrass – AWS Edge Computing Platform (2)

▪ Greengrass provides connector, connecting edge-fog- cloud nodes, and realizing adaptive configuration

slide-65
SLIDE 65

Greengrass – AWS Edge Computing Platform (3)

▪ Greengrass provides good authorization and privacy protection mechanisms

slide-66
SLIDE 66

Greengrass – AWS Edge Computing Platform (4)

slide-67
SLIDE 67

AWS IoT Core – Edge node (1)

slide-68
SLIDE 68

AWS IoT Core – Edge node (2)

slide-69
SLIDE 69

Greengrass → IoT Analytics

slide-70
SLIDE 70

AWS IoT Analytics (1)

slide-71
SLIDE 71

AWS IoT Analytics (2)

slide-72
SLIDE 72

AWS Lambda (1)

▪ AWS lambda is a fine-grained method for deploying code, managing services, and monitoring the health of lightweight services. similar to Alibaba microservice.

slide-73
SLIDE 73

AWS Lambda (2)

▪ AWS lambda is a new pricing and service model

slide-74
SLIDE 74

AWS Lambda (3)

slide-75
SLIDE 75

Microsoft IoT Solution

Simple processing

filtering, batching, compression

Complex processing

Azure Stream Analytics, Cognitive Services

slide-76
SLIDE 76

Microsoft IoT Core Innovations (1)

slide-77
SLIDE 77

Microsoft IoT Core Innovations (2)

slide-78
SLIDE 78

Microsoft IoT Processing Procedure

slide-79
SLIDE 79

Microsoft IoT Intelligent Processing Lambda Architecture

slide-80
SLIDE 80

Microsoft IoT Connection Procedure

slide-81
SLIDE 81

Microsoft IoT Edge-Fog-Cloud Fusion (1)

slide-82
SLIDE 82

Microsoft IoT Edge-Fog-Cloud Fusion (2)

slide-83
SLIDE 83

Thank you!