ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 ›...

15
3/20/2015 Copyright © 2015, SAS Institute Inc. All rights reserved. 1 Copyright © 2015, SAS Institute Inc. All rights reserved. ALL THINGS DIGITAL MATHIAS COOPMANS Copyright © 2015, SAS Institute Inc. All rights reserved.

Transcript of ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 ›...

Page 1: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 1

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

ALL THINGS DIGITALMATHIAS COOPMANS

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Page 2: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 2

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Copyr i gh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Page 3: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 3

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Copyr i gh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Page 4: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 4

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Copyr i gh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Page 5: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 5

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

CONNECTED TRUCK KEEP ON ROLLING

Business Goal

• Predict maintenance needs of individual trucks

before failures occur

• Proactively service trucks at opportune time

• Provide new service offering with high SLA

Image credit: Mike, https://www.flickr.com/people/pmiaki/

Pilot Project Results

• Models able to predict likely failures 30 days in advance with 90% accuracy

• Better root cause insight led to smaller campaign

Process:• Data from 60+ sensors/truck.• Integrate the data with product

details, warranty claims, and related data sources.

• Build analytic models to predict the likelihood of specific failures within 30 days.

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Page 6: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 6

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

-2 hoursA1

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

PREDICTION ACCURACY HOW ACCURATE ARE THE PREDICTION MODELS?

ProcessTripAccuracy

1h beforeAccuracy2h before

80%65%

Accuracy2h before

60%

Feeder data

Feeder + Extruder data

Page 7: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

Slide 11

A1 Author, 11/12/2013

Page 8: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 7

Copyr igh t © 2012 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

3 ENABLING TECHNOLOGIES TO FOLLOW

SAS® Event Stream

Processing Engine

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

INTRODUCTION WHAT IS SAS® EVENT STREAM PROCESSING ?

ENGINE

DATA IN (called Events)

DATA OUT(Events)

Process DataOn the Move

Very High speedLow latency

Page 9: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 8

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

INTRODUCTION DATA PROCESSING ON THE MOVE

BATCH ENGINE STREAM ENGINE

1. Prepare data2. Run Process3. Get results4. Go to step 1

1. Run Process2. Continuous loop :

a. Receive data inb. Process datac. Push results out

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

INTRODUCTION DATA PROCESSING AT HIGH SPEED

SAS EVENT STREAM PROCESSING ENGINE

DATA IN (called Events)

DATA OUT(Events)

From huge volume of streaming data flowing at

very high rate : Millions of records/sec

Data (Events) are processed as soon as they

arrive (happen) :Latency: < 1 millisecond

Page 10: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 9

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

INTRODUCTION DATA PROCESSING ON THE MOVE AND AT HIGH SPEED

SAS EVENT STREAM PROCESSING ENGINE

DATA IN (called Events)

DATA OUT(Events)

Detect Events of Interest

FilteringAggregation

Pattern detectionCalculationsCorrelationsProcedural

Thresholding

and much more…

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

INTRODUCTION DATA PROCESSING ON THE MOVE AND ATHIGH SPEED

SAS EVENT STREAM PROCESSING ENGINE

DATA IN (called Events)

DATA OUT(Events)

Detect Events of Interest

FilteringAggregation

Pattern detectionCalculationsCorrelationsProcedural

Thresholding

and much more…

TYPICAL ESP QUESTIONS

“ give me the top 3 values every 5 minutes”

“Tell me when an event A was followed by an event B and not event C within 3 minutes”

“Tell me when you detect a 3rd bank transfer in the last 24 hours from the same account, coming from 3 different countries, and pause the transfer until manual validation”

“Filter out sensor readings when the device was in maintenance period”

Page 11: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 10

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

KEY CONCEPTS DATAFLOW CENTRIC

SAS EVENT STREAM PROCESSING ENGINE

DATA IN (Events)

DATA OUT(Events)

Event Stream

Event Stream

Event StreamEvent

Stream

Event Stream

Event Stream

Event Stream

Design of the rule model (called “Continuous Query”)using components (called “Windows”)

Event Stream

DATA IN (Events)

DATA IN (Events)

DATA OUT(Events)

SOURCE1

WINDOW

SOURCE2

WINDOW

SOURCE3

WINDOW

FILTER

WINDOW

CALCULATIONS

WINDOW

JOIN

WINDOW

JOIN

WINDOW

CALCULATIONS

WINDOW

THRESHOLD

WINDOW

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

AN EXPLANATION…SAS IN-MEMORYWhether a personal computer or an enterprise server, a computer is made up of three essential components:

MEMORY

STORAGE

PROCESSING

DiskDiskDisk

RAM

CPU CPU

Page 12: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 11

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

A HIGH-LEVEL OVERVIEW OF OUR

APPROACH

A.Data In-Memory• Allows for ultra-fast access to the data

B.Extreme Parallelism• Fully leverages all processor cores

C.Distribution of Analytics Processes• Fully exploiting all resources of multiple blades/nodes:

RAM and Processor cores

RAM

CPU CPU

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

Data NodesHead

HADOOP FUNDAMENTALS

Page 13: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 12

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

2 KEY FEATURES DISTRIBUTED STORAGE, DISTRIBUTED PROCESSING

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

HDFS DATA BLOCKED, DISTRIBUTED & REPLICATED

Page 14: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 13

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

OVERVIEW FROM, WITH AND IN HADOOP

SAS can treat Hadoop just as any other data source, pulling data

FROM Hadoop, when it is most convenient;

SAS can work WITH Hadoop, lifting data in a purpose-built

advanced analytics in-memory environment;

SAS can work directly IN Hadoop, leveraging the distributed

processing capabilities of Hadoop.

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

FROM + WITH + IN HADOOP IS NOT AN OR, BUT AN AND

Prepare data INHadoop for analytics

Move it FROM Hadoop into a SAS server

Deploy and manage model score code IN Hadoop

HPA temporarily lifts data IN to memory for analytics at scale

Model data at scale in-memory WITH visual statistics and in-memory statistics

TIPUse the right technique for what needs to

be done!

Explore data at scale, in-memory WITH visual analytics

Page 15: ALL THINGS DIGITAL - Sas Institute › content › dam › SAS › en_be › doc › other2 › ...“ give me the top 3 values every 5 minutes” “Tell me when an event A was followed

3/20/2015

Copyr i ght © 2015 , SAS I ns t i tu te I nc . Al l r i gh ts reserved . 14

Copyr igh t © 2015 , SAS Ins t i t u t e Inc . A l l r i gh t s res e rv ed .

THANKS FOR YOUR ATTENTION