Fog Computing as a foundational architectural concept for Internet of Things (IoT) and Internet of Everything (IoE).
Embedded devices in the IoT are hampered by the compute, storage, and service limitations of living life on the edge. As IoT edge devices comprise broader sensor networks for industrial automation, transportation, and other safety critical applications, their high uptime requirements are nonnegotiable and service latencies must be kept within realtime or near real time parameters. However, the size, weight, power, and cost constraints of edge platforms also inhibit the ondevice resources available for executing such functions. In this session, Gandhi will introduce Fog Computing, a new paradigm for the IoT that extends compute, storage, and application resources from the cloud to the network edge. Beyond the interplay between Fog and Cloud, Gandhi will show how Fog services can be leveraged across a range of heterogeneous platforms—from end user devices and access points to edge routers and switches—through software technology that facilitates the collection, storage, analysis, and fusion of data to drive success in your next IoT device deployment.
Get Cloud Resources to the IoT Edge with Fog Computing
2. Get
Cloud
Resources
to
the
IoT
Edge
with
Fog
Compu8ng
Biren
Gandhi
Principal
Strategist
Corporate
Strategic
Innova8on
Group,
Cisco
@birengandhi
/in/birengandhi
3. Abstract:
Embedded
devices
in
the
IoT
are
hampered
by
the
compute,
storage,
and
service
limita8ons
of
living
life
on
the
edge.
As
IoT
edge
devices
comprise
broader
sensor
networks
for
industrial
automa8on,
transporta8on,
and
other
safety
cri8cal
applica8ons,
their
high
up8me
requirements
are
nonnego8able
and
service
latencies
must
be
kept
within
real-‐8me
or
near
real-‐8me
parameters.
However,
the
size,
weight,
power,
and
cost
constraints
of
edge
plaLorms
also
inhibit
the
on
device
resources
available
for
execu8ng
such
func8ons.
In
this
session,
Gandhi
will
introduce
Fog
Compu8ng,
a
new
paradigm
for
the
IoT
that
extends
compute,
storage,
and
applica8on
resources
from
the
cloud
to
the
network
edge.
Beyond
the
interplay
between
Fog
and
Cloud,
Gandhi
will
show
how
Fog
services
can
be
leveraged
across
a
range
of
heterogeneous
plaLorms—from
end
user
devices
and
access
points
to
edge
routers
and
switches—through
soOware
technology
that
facilitates
the
collec8on,
storage,
analysis,
and
fusion
of
data
to
drive
success
in
your
next
IoT
device
deployment.
Simplified
Abstract:
What
is
Fog
Compu8ng
and
how
can
it
solve
IoT
pain
points?
4. IoT
(and
IoE)
Delivering
the
Right
Informa8on
to
the
Right
Person
(or
Machine)
at
the
Right
Time
Process
Physical
Devices
and
Objects
Connected
to
the
Internet
and
Each
Other
for
Intelligent
Decision
Making
Things
Connec8ng
People
in
More
Relevant,
Valuable
Ways
People
Leveraging
Data
into
More
Useful
Informa8on
for
Decision
Making
Data
IoE
Networked
Connec8on
of
People,
Process,
Data,
Things
5. IoT
is
Here
–
Now
and
Growing
TIMELINE
2010
2015
2020
BILLIONS
OF
DEVICES
0
10
20
30
50
Adop?on
rate
of
digital
infrastructure:
5X
faster
than
electricity
and
telephony
25
12.5
7.2
6.8
7.6
World
Popula?on
50
Billion
Smart
Objects
Inflec?on
point
6. IoT
Ver8cals
in
Ac8on
Source:
Cisco
IoT
Purchase
Process
Global
Study
January
2015,
N=
2582
MANUFACTURING
Inventory
Management
Real-‐8me
Monitoring
Energy
Management
GOVT
TheO
Protec8on
Asset
tracking
Real-‐8me
Billing
UTILITIES
Fleet
Management
Real-‐8me
Asset
Tracking
Safety
and
Compliance
TRANSPORTATION
Fleet
Management
Inventory
Management
Op8mize
delivery
RETAIL
TheO
Protec8on
Inventory
Management
Real-‐8me
Billing
7. Delivering
Outcomes
-‐
NOW
MANUFACTURING
Safety
Product
Quality
Supply
Cain
and
Logis8cs
GOVT
Safety
Efficiency
Customer
Sa8sfac8on
UTILITIES
Safety
Customer
Sa8sfac8on
Supply
Chain
and
Logis8cs
TRANSPORTATION
Safety
Customer
Sa8sfac8on
Supply
Chain
and
Logis8cs
RETAIL
Safety
Customer
Sa8sfac8on
Cost
Reduc8on
Source:
Cisco
IoT
Purchase
Process
Global
Study
January
2015,
N=
2582
9. And
So
Do
Challenges
…
Many
of
these
are
technology
challenges
10. Key
IoT
Challenges
(1)
• Data
Size/Sources
(Velocity,
Variety,
Volume)
– Geo-‐distribu8on
– M2M
Chamer
(status,
health,
etc.)
• IT
meets
OT
– Local
Control
Loops
(determinis8c
behavior)
– Cyber
Physical
Threats/Security
• System-‐wide
View
– e.g.
Smart
Traffic
Light
System
11. Key
IoT
Challenges
(2)
• Cohesive
Opera8ons
– Resource
Orchestra8on
&
Monitoring
– Distributed
Policy
Management
– Joint/Delegated/Contractual
Ownership
• Data
Processing
&
Analy8cs
– Balance
among
Real-‐8me,
Semi-‐real-‐8me
and
Non-‐real-‐8me
– Aggrega8on
and
Bandwidth/Compute
Cost
12. Key
IoT
Challenges
(3)
• Reliability/Availability
– Unlike
DC,
“failure”
is
a
norm
• Complex
Greenfield/Brownfield
Deployment
– Longer
Life
of
Cri8cal
Infrastructural
Systems
– Huge
Deployment/Opera8onal
Costs
• Interplay
with
the
Cloud
– Many
Silo’ed
“PlaLorms”
(per
vendor)
– Detached
Apps
(driven
from
consumer
domain)
– Ecosystem
of
Tools
and
Prac8ces
13. In
Summary
IoT
provides
huge,
disrup8ve
opportuni8es
IoT
is
rapidly
adopted
at
an
accelerated
pace
IoT
demands
new
architectural
paradigm
to
deal
with
unique
challenges
14. Meet
“Fog
Compu8ng”
“A
distributed
compu8ng
paradigm
that
extends
cloud
compu.ng
closer
to
the
edge
of
the
network
to
enable
new
wave
of
applica.ons
and
services
in
IoT
land”
15. Example:
Smart
Traffic
Light
System
• Traffic
Light
at
Intersec8on
– Goal:
accident
preven.on
– Fair
Traffic
Management
– Detects
vehicles/
pedestrians/cyclist
crossing
– Measures
distance
&
speed
of
approaching
vehicles
– Issues
alarms
to
approaching
vehicles
– Changes
from
Green
to
Red
– Takes
photos
&
issues
8cket
• STLS
as
a
system
– Goal:
facilitate
smooth
traffic
flow
throughout
the
city/region
– Traffic
conges8on
measurements
– Route
recommenda8on
(no
loops,
please!)
– Coordina8on
of
cycles
of
individual
intersec8on
– Emergency
Evacua8on
Assistance
– Crime-‐in-‐progress
Detec8on
16. Simplified
Requirements
• Local
subsystem
(traffic
light,
sensors,
actuators)
-‐ Reac8on
8me
<
10
msec
-‐ Compute/storage
capability
-‐ Ruggedized
&
small
form
factor
box
• Global
system
-‐ Wide
geo-‐distribu8on
-‐ Middleware
orchestra8on
-‐ Mul8plicity
of
agencies
running
the
system
(must
coordinate
control
policies)
• Interac8on
with
Cloud
– Efficient
traffic
management
demands
large
database
of
historical
records
w/
heuris8cs/analy8cs
capabili8es
This
is
Fog
Compu8ng!
(cloud
–
but
closer
to
the
ground)
17. What
Exactly
is
Fog?
INTERNET
OF
THINGS
APPLICATIONS
Cloud
Fog
Cloud
capabili8es
extended
closer
to
the
edge
Consumer
domain’s
app-‐
centric
model
of
device-‐to-‐
cloud
may
not
be
enough
18. Hierarchical
Fog
Architecture
Cloud
Device/Smart
Object
East/West
Flows
Fog
Fog Nodes can be multi-tenant
Shared, public or private (like cloud)
Highly virtualized environment
Secured & isolated tenants, QoS, workload distribution
Service Mobility
Ability to migrate a running instance from cloud to edge
North/South
Flows
Mixed ownership & operation
Single entity, federation of agencies
19. Fog
Compu8ng
PlaLorm
• Enables
low-‐latency,
near-‐real-‐8me
compute
capabili8es
closer
to
the
edge
• Tames
data
deluge
and
effec8vely
u8lizes
costly
bandwidth
• Leverages
proven
cloud
eco-‐system
of
tools,
technologies,
best
prac8ces
and
developer
API
to
support
mul8-‐agency
orchestra8on
and
management
In
short,
think
of
it
as
a
Sophis8cated
App
Store
model
for
heterogeneous
IoT
infrastructure
20. Fog
Compu8ng
Example
Use
Cases
Smart Traffic Lights
Real-time (RT) local control loop
Geo-distributed orchestration
Multiagency policy co-ordination
Local/Global Analytics
L
G
C
O
Wind Farm
RT local control loop
In-situ orchestration
Global Big Data
L
G
C
Connected Rail
Two-tier wireless AP
Fast mobility
Low latency streaming
RT actionable analytics
Global big data
M
L
G
C
O
Retailing
Video analytics
Interplay between local and
Globally processed data
L
C
Oil & Gas
RT actionable analytics
Geo-distributed Orchestration
Industrial automation, Big data
M
L
G
C
SCV & Transport
RT actionable analytics
Global Big Data
(batch processing)
M
L
C
Military Apps
Real-time local control loop
Geo-distributed Orchestration
Multiagency policy co-ordination
Local/Global Analytics
M
L
G
C
O
M
L
G
C
O
Mobility
Geo-‐distribu8on
Low/predictable
latency
Cloud
interac8on
Mul8-‐agent
orchestra8on
Cri8cal
amributes
21. Data
is
the
new
$$$
1.1 Billion
Data points generated by sensors daily
500 Gigabytes
Data generated by an offshore oil rig weekly
1000 Gigabytes
Data generated by an oil refinery daily
10,000 Gigabytes
Data generated by a jet engine every 30 minutes
2.5 Billion Gigabytes
Data generated worldwide daily
90% of the world’s data has been created in the last 2 years!
IoT
Brings
New
Data
Dimensions
22. Data
Interplay
between
Fog
&
Cloud
FogSensors Cloud
Storage
Realtime Data
Control/Actuation
Transient
milliSec/Seconds
Visualisation
_-‐-‐_-‐
Analytics
Seconds/Minutes
Semi-Permanent Months/Years
Minutes/Days/Weeks
Filter to process data locally
Remaining pass to cloud
Response
_-‐-‐_-‐
Business Intelligence
Dashboards
KPIs
_-‐-‐_-‐
_-‐-‐_-‐
Control
Loop
In
8me
and
in
space
Localization Globalization
23. Data
Processing
at
the
Edge
• Need
of
immediate
response
8me
– Can't
afford
latency
of
sending
up
and
back
the
chain
• Closed-‐loop
control
– In
controlling
physical
systems
–
cannot
depend
on
speed
and
availability
of
resources
back
at
the
data
center
–
e.g.
smart
traffic
light
system
• Privacy,
Security
&
Data-‐ownership
considera8ons
– Regulatory
and
business
concerns
may
not
allow
moving
the
data
• Improved
scale
and
aggregate
throughput
via
parallelism
– Data
sources
oOen
naturally
distributed
• Not
all
Bytes
are
born
equal
– Offload
centralized
resources
that
would
otherwise
have
to
filter
through
volumes
of
uninteres8ng/useless
data.
24. Fog
Compu8ng
–
Buzz
or
Business?
Let’s
check
out
some
use
cases
25. Railway
Systems
Immediate
Response
to
Equipment
Failure
Real-‐8me
Health
Status
of
Trains
&
Tracks
New
Passenger
Ameni8es
and
Services
REPLACE
BEARINGS
CAR
07
26. Oil
Pipelines
Proac8ve
Leak
Detec8on
Predic8ve
Management
Broadband
Connec8on
for
Community
S6 | C026 Pressure
Drop
kPa
ACTION
REQUIRED
27. U8lity
Substa8on
Local
Load
Balancing
Opera8onal
Efficiencies
New
Business
Models
01 02 03 04 05 06 07 08 09 10
Maintenance
Needed
on
Transformer
7
28. Manufacturing
Predic8ve
Maintenance
Op8mize
Energy
Usage
Security
Before,
During,
and
AOer
Amack
QUALITY
POWER
STAMPS
35%
85%
592547/
1000000
30. Fog
Compu8ng
@
Cisco
Live
Video
of
Fog
Compu8ng
demo
at
Cisco
Live
San
Diego
31. Cisco
IoT
System
INTERNET
OF
THINGS
APPLICATIONS
Fog
Compu?ng
Management
and
Automa?on
Network
Connec?vity
Security
Cyber
and
Physical
Data
Analy?cs
Applica?on
Enablement
PlaWorm
Cloud
Fog