iotand edge computing - stanford university€¦ · data service edge cloud ingress real time...
TRANSCRIPT
IoT and Edge Computing
Satyam VaghaniVP & GM, IOT & AI
Where is everybody?
N O T P U B L I C C L O U D P U B L I C C L O U D
Private DCs 201732 million nodes
Top-3 providers 20173.5 million nodes
Enterprise IoT 20173.135 billion devices
Enterprise IoT 20207.5 billion devices
Sources: Gartner, IDC
A New Paradigm Needs a New OS
HUMANS DATACENTERS THINGS
DESKTOPWindowsMac OS
MOBILEiOS
Android
PRIVATE CLOUDvSphereAcropolis
PUBLIC CLOUDAWSGCE
?
No, Not That OS
Scope of an IoT OS
2017 2020
Cloud traffic
8.6 ZB 15.3 ZB
2017 2020
IoT data
256 ZB
600 ZB
Key Change: Edge Trumps Cloud
Sources: Cisco Global Cloud Index, Memoori
AFTER
AFTER
BEFORE
Data IngestionReal-time
ProcessingLong-termProcessing
Real-timeProcessing
Long-termProcessingData Ingestion
Key Consequence: The Intelligent Edge
SENSORS
SENSORS
CLOUD
CLOUD
EDGE COMPUTING
IoT GATEWAY
The Intelligent Edge is Not Ready
TWO PROBLEMS PREVENT WIDESPREAD ADOPTION• Distributed infrastructure burden• Domain expertise burden
A Smart Airport ExampleTOPOLOGY
Airport 1 (of 10)
2000 x
100 x
250 x
A Smart Airport Use CaseOBJECT OF INTERESTLook for a red car at airport(s)
SFO AIRPORT (EDGE) CLOUD
...10:41AM: redcar=010:42AM: redcar=110:43AM: redcar=010:44AM: redcar=0...
...10:41AM: SFO, redcar=010:42AM: SFO, redcar=110:43AM: SFO, redcar=010:44AM: SFO, redcar=0...
EXPLICIT DISTRIBUTED SYSTEMS, MACHINE LEARNING, IMAGE RECOGNITION, SECURITY, STORAGE, NETWORK PROGRAMMING, AND MORE
IoT Infrastructure Burden: Apps Span Edge and Cloud
Train model to recognize red car
Deploy model to selected edges
Runtime for car recognition model
Persistence for surveillance feed
Sampling surveillance feed to match model
input requirements
Persistence of image recognition time series output Data mover to
move time series output to cloud
Persistence for time series data in cloud
App runtime in cloud
Security
business logic
infrastructure madness
IoT Development Burden: Domain Knowledge
IoT APP CREATION REQUIRES DOMAIN EXPERTISE IN MULTIPLE AREAS
Analytics Machine Learning Security
Data Pipelining Cloud Services …etc.
Don’t try this at home
Machine Learning
Reduce infrastructureburden
Reduce developmentburden
An Ideal IoT “OS”: Hyperconvergence 3.0
EDGE CLOUD
Sensor services
Data services
App services
App services
Data services
ENABLE RAPID DEVELOPMENT OF IoT APPS
MAKE IoT INFRASTRUCTURE SEAMLESS + INVISIBLE
2
1
DemoHIGHLIGHTS
• Multi-tenant application runtime across Edge and Cloud• 1-click data pipelines across Edge and Cloud
MAKE IoT INFRASTRUCTURE SEAMLESS + INVISIBLE 1
• Modular functions for rapid development of IoT apps
ENABLE RAPID DEVELOPMENT OF IoT APPS2
Global time
series in cloud
Time
series
Modular Functions for Rapid Development of IoT Apps
FUNCTIONDatastream
in
Datastream
out
Parameters
VIDEOPROCESSING
OBJECTRECOGNITION
DATAMOVER
Image
stream
Live video
sample, crop redcar, presence Google, structured
person, count
Persistence
Runtime
High level IoT OS Architecture
Data Ingestion
Sensor Actuation
Real-time Data Service
Unstructured Data Service
Structured Data Service
Data Bus
Machine Training
Streaming Analytics
Custom Code
Control Plane
Operations Console
Developer Console
SENSORS
Machine Training
Long term Analytics
Custom Code
Streaming Data Service
Unstructured Data Service
Structured Data Service
EDGE CLOUD
LONG TERM PROCESSINGREAL TIME PROCESSINGINGRESS
MUCH BIGGER THAN IoT
Persistence
Runtime
The Long View: A Data and Code Hypervisor for Hybrid Apps
Data Ingestion
Data Archival
Streaming Data Service
Unstructured Data Service
Structured Data Service
Data Bus
Machine Learning
Streaming Analytics
Custom Code
Control Plane
Operations Console
Developer Console
DATA LAKE
Machine Training
Long term Analytics
Custom Code
Streaming Data Service
Unstructured Data Service
Structured Data Service
LOCAL CLOUD OTHER CLOUDS
REMOTE PROCESSINGLOCAL PROCESSINGINGRESS
Use casesA RANDOM WALK• Your doctor’s office• At Great Mall• The airline. And the government. And an airport authority• ONG services• The car company• The county
GTM experiences
Thank YouSatyam Vaghani | @SatyamVaghani