►
From YouTube: Webinar: Automation, AI and ML for Network Management: It’s Already Here, Are Organizations Ready?
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
From physical to virtual to cloud (and now multi-cloud) infrastructures, networks are getting more diverse and challenging to manage. Humans and manual processes can no longer keep pace with network innovation, evolution, complexity, and change.
Organizations are adapting, through automation, AI and ML.
In this session, Kentik’s VP of Marketing Andy Singer, shares the feedback his team collected from more than 380 technical and executive industry peers regarding adoption and readiness for automation, AI and ML for network management. Learn what you need to get ready now and how your organization stacks up.
A
All
right,
I
think
we're
gonna
go
ahead
and
get
started
for
a
couple
minutes
into
the
hour
looks
like
we
have
a
good
group
of
folks
here
so
I'd
like
to
thank
everyone
joining
us.
Welcome
to
today's
CN
CF
webinar,
which
is
a
bit
of
a
mouthful
but
automation,
AI
and
ml
for
network
management.
It's
already
here
our
organization's
ready.
A
A
If
you
post
a
question
in
chat,
but
that
way
we
can
actually
track
questions
and
make
sure
they
get
answered
so
feel
free
to
drop
questions
and
there
this
is
an
official
webinar
of
the
CN
CF
and,
as
such,
it's
subject
to
the
CN
CF
code
of
conduct,
please
don't
add
anything
to
the
chat
or
the
Q&A.
That
would
be
a
violation
of
that
code.
A
B
You
Phil,
just
making
sure
you
guys
can
hear
me.
Okay,
I
think
so
looks
good
all
right
good
morning,
good
afternoon
and
good
evening.
Everyone
and
thank
you
for
joining
today's
webinar
Phil
I'm
I'm
Ken
dick,
is
honored
to
be
able
to
be
able
to
present
to
this
audience,
and
we
appreciate
you
moderate
moderating
today's
sessions
or
session.
If
anybody
has
questions
during
today's
presentation
feel
free
to
ask
Phil
will
moderate
and
I
will
do
my
best
to
answer
them
and
if
I
can't,
we
will
get
you
a
response
after
the
webinar.
B
B
So,
first,
a
little
bit
about
me
and
Ken
t'k,
so
I
lead
the
marketing
function
here
at
Kent,
ik
Ken
tech
provides
revolutionary
insights
and
automation
to
make
every
network
excellent.
We
power
up
operations,
teams
with
AI
ops
for
network
professionals.
We
were
founded
in
2014,
our
headquarter
is
based
in
San
Francisco
and
we
have
employees
all
over
the
world
and
we
have
over
250
customers.
B
Today,
it's
very
much
a
cloud
centric
world
and
I
think
we
all
know
that,
just
by
nature
that
we're
on
a
CN
CF
webinar,
but
it's
an
important
thing
to
call
out
that
the
cloud
is
really
at
the
center
of
everything
that
happens
today
with
digital
businesses
and
most
businesses
in
general
and
for
consumers.
We
know
that
all
this
data
is
flowing
to
and
from
different
clouds
to
and
from
different
cloud
based
applications.
B
Many
enterprises
and
businesses
and
providers
have
lots
of
data
centers,
whether
newer,
software-defined
or
legacy
data
centers.
In
addition,
companies
and
enterprises
and
organizations
still
have
and
rely
on
headquarters
and
branch
offices,
all
that
connectivity
flows
to
and
through
the
internet
and
to
and
through
and
from
the
cloud.
B
We
know
that
networks
are
changing
and,
in
my
experience
in
the
last
you
know,
20
25
years,
I've
seen
tremendous
change
in
networks,
but
I
think
some
of
the
most
transformative
change
has
come
in
the
last
10
years
and
we've
seen
the
rise
of
the
cloud
and
what
that's
done
for
networking.
So
now
today
we
have
multiple
architectures
in
place.
Physical
virtual
we've
got
increased
traffic
complexity.
B
We've
got
a
ton
of
API
driven
use
cases,
so
the
rise
of
API
interconnectivity
is
really
driving
a
lot
of
the
explosive
growth
that's
taking
place
today,
both
in
networks
and
in
applications.
We've
got
an
increasing
amount
of
automation
and
we're
going
to
talk
about
that.
But
when
we
talk
about
automation,
we're
specifically
talking
about
manual
tasks
that
can
be
automated
to
drive,
increase
efficiency
and
efficacy.
B
B
We
know
in
all
of
our
conversations
with
our
customers
and
prospects
and
the
market
that
we
engage,
that
network
teams,
network
professionals
all
up
and
down
the
market.
Small
companies,
medium
companies,
large
companies,
are
struggling
to
some
degree.
We
know
that
network
outages
are
becoming
a
problem
largely
due
to
the
complexity.
That's
there
today
with
networks,
and
a
recent
Amazon
outage
was
estimated
to
cost
about
99
million
dollars
in
lost
revenue.
B
Southwest
Airlines
network
outage
costs
had
the
potential
to
reach
approximately
82
million
dollars
and
when
we
think
about,
and
when
we
talk
to
the
people
that
are
thinking
about
outages
every
day,
here
are
some
of
the
things
that
bubble
up:
here's
what
they
tell
us,
they
say
number
one.
We
need
to
be
able
to
quickly
detect
problems,
and
we
know
it's
difficult
because
of
the
complexity
of
the
networks
that
we
have
today,
whether
we're
a
hundred
percent
cloud,
multi
cloud
or
even
hybrid
cloud,
there's
a
huge
need
to
be
able
to
quickly
detect
problems.
B
We
know
that
performing
root.
Cause
analysis
is
even
more
complicated
just
by
nature.
The
fact
that
there's
so
many
different
types
of
data
coming
in
that
you
need
to
be
able
to
analyze
and
need
to
be
able
to
look
back
at
data
historically
to
be
able
to
really
understand
the
root
cause
of
a
network
problem.
B
Migrating
applications
is
becoming
increasingly
complicated,
not
just
because
it's
hard
to
begin
with,
but
because
of
things
like
different
architectures,
whether
you're
using
containers
or
not,
whether
you
have
single
cloud
or
multi
cloud,
multiple
operating
systems
in
place,
even
multiple
hypervisors
and
virtualization
platforms.
Migrating
applications
is
a
big
challenge
planning
for
capacities.
Another
one
capacity
is
really
important
because
of
the
demand
out
there
for
applications
in
the
cloud
you
need
to
be
able
to
understand.
Do
you
have
enough
capacity?
Where
are
you?
Gonna
need
new
capacity?
B
Where
are
you
exceeding
your
playing
capacity
and
are
you
paying
too
much
for
that
capacity?
And
then
that
leads
me
to
allocating
network
costs
more
and
more
as
different
types
of
transit
are
out
there
on
different
types
of
networks
are
in
place.
Network
professionals
have
the
ability
to
play
a
bigger
role
in
helping
businesses,
maintain
and
manage
the
cost
for
network.
B
So
I
want
to
now
transition
into
a
survey
that
we
conducted
called
the
state
of
automation,
artificial
intelligence
and
machine
learning
and
network
management.
So
we
went
to
Cisco
live
earlier.
This
year
we
executed
our
survey
and
we
got
388
conference
attendees
to
give
us
really
valuable
data
on
everything
from
their
business
and
industry
to
their
thoughts
on
several
key
questions
around
this
specific
topic,
the
respondents
were
had
lots
of
different
titles.
We
got
c-level
folks
s
VP's
bps,
but
a
majority
of
respondent
reported
having
a
title
of
network
engineer.
B
B
Network
automation
is
taking
shape
and
will
provide
some
data
that
talks
about
how
the
types
of
automation
that
are
being
used
goes
across
many
different
categories
and
even
taking
place
more
and
more
on
network
management
tasks.
The
third
finding
is
networking
processes
like
compliance
and
Incident
Response
are
less
likely
to
be
automated.
So
when
it
comes
to
things
that
have
to
do
with
regulatory
issues
or
even
security,
those
tasks
are
still
being
performed
today
by
people
and
not
being
automated,
and
finally,
machine
learning
is
growing
in
importance
for
network
management.
B
More
and
more
people
are
realizing
that
the
complexity
of
today's
networks
and
the
speed
at
which
we
operate
in
the
real
time
nature
of
all
of
our
businesses
requires
more
and
more
machine
learning
to
be
used
to
be
able
to
help
everybody
better
manage
their
networks,
and
we,
this
was
true
across
the
board
industry
title,
regardless
of
who
you
ask
Phil
I'm
gonna
pause
right
there
just
for
a
quick
second.
Do
we
have
any
questions
in
the
queue
a
we
don't
yet
at.
B
Some
additional
data
are
are
includes,
for
example,
the
greatest
adopters
of
cloud
services
are
the
education
sector,
but
the
vast
majority
of
respondents
said
that
50%
arghhhh
or
more
use
only
one
cloud
technology
and
Finance
sectors
gave
the
strongest
responses
confirming
multi-cloud
use
at
43%
38%,
and
the
energy
sector
was
a
close
third
at
37%.
In
addition,
you
know
we
think
about
these
results.
Well,
our
takeaway
is
that
networking
pros
are
already
dealing
with
significant
operational
complexity.
It's
it's
hard
enough
to
move
from
a
physical
to
a
virtualized
environment.
B
B
I
think
from
our
report,
we've
got
a
couple
different
stats.
We've
got
some
things
by
vertical,
showing
that
how
they've
moved
to
the
cloud
and
actually
a
great
way
to
get
you
this
information
is
to
actually
get
you
the
report
which
has
actually
cloud
adoption,
single
and
multi
cloud
by
industry.
So
what
we'll
do
is
we'll
put
the
link
in
the
QA
to
be
able
to
download
the
report
and
that
will
better
answer.
Your
question.
B
So
our
second
key
finding
is
around
network
automation.
It's
definitely
taking
shape.
So
what
we
asked
are
that
what
we
asked
respondents
is
how
ready
is
your
organization
for
full
automation
to
manage
network
performance
and
network
security?
Our
takeaway
is
that
85%
of
respondents
reported
having
at
least
one
Network
automation
deployment
in
place.
55
percent
reported
having
two
or
more
deployments,
and
27
percent
of
respondents
reported
that
their
organization
is
either
extremely
prepared
or
very
prepared
for
full
automation.
B
Now
this,
in
my
view,
is
actually
really
exciting,
because,
even
as
long
as
just
last
year
or
even
two
years
ago,
network
automation
was
kind
of
largely
shunned
as
something
that's,
maybe
frightening
or
unreliable,
but
that's
changed.
The
more
and
more
companies
are
adopting
software-defined
networking
automation
takes
a
bigger
and
bigger
role,
but
IT
broader
I,
see.
Organizations
in
general
are
no
longer
willing
to
wait
for
networks
to
be
manually
configured.
So
when
it
comes
to
cloud,
we
know
that
things
like
provisioning
bursting.
B
These
are
all
things
that
happen
automatically
and
a
lot
of
the
networking
in
there
is
automatically
provisioned,
but
when
it
comes
to
network
management,
troubleshooting
and
remediation,
these
are
the
things
where
automation
is
taking
the
next
big
leap.
You
know
when
we
look
back
at
last
year.
We
did
the
same
survey
and
same
question
and
last
year's
survey,
and
we
showed
that
only
15%
of
respondents
felt
their
organization
was
prepared
for
automation.
So
we've
had
a
huge
increase
in
the
number
of
companies,
saying
yeah
we're
ready
for
automation.
B
B
B
When
we
asked
the
respondents
kind
of
what
are
they
using
automation,
for
these
are
the
top
categories:
cloudbursting
configuration
compliance
incident
response,
policy
management
and
workload
management,
a
majority
of
respondents,
53%
are
deploying
automation
for
configuration
purposes
and
that
makes
total
sense.
There's
a
ton
of
manual
effort
required
in
doing
configuration
or
around
networking,
and
it
makes
sense
to
automate
those
tasks.
B
40%
reported
using
automation
for
policy
management,
and
this
is
actually
a
very
interesting
stat.
More
and
more
automation
is
being
used
for
policy
configuration
so
whether
you're
getting
you're
creating
policy
on
a
daily
basis
or
you're
modifying
policy
in
real
time.
Let's
say:
you're
moving
a
network
interface
from
A
to
B
or
you're
configuring
different
routes.
All
those
policy
changes
are
becoming
increasingly
automated.
B
You
know
what
we
think
about
this
is
that
the
whole
point
of
cloud
bursting
is
to
do
so
an
automated
and
dynamic
basis.
Our
thought
is
that
the
networking
components
to
enable
cloud
bursting
are
most
likely
already
in
place
and
don't
need
to
change
so
cloud
bursting
is
a
special
enough
use
case.
That's
not
common!
Amongst
the
you
know,
all
these
respondents
now
there's
two
ways
to
look
at
it.
There's
one
way
people
could
just
say
it's
already
automated
in
my
virtualization
platform.
Therefore
I
don't
need
to
worry
about
it.
B
B
Then,
when
we
get
into
machine
learning,
we
know
that
machine
learning
is
growing
in
importance
for
network
management.
So
when
we
take
a
look
at,
we
asked
everybody
their
stance
on
machine
learning
and
over
whelming.
Sixty-Five
percent
of
respondents
said
the
machine
learning
is
now
extremely
are
very
important
for
network
management.
A
B
A
Yeah,
so
it's
a
little
bit
long,
I'll
read
it.
What
are
the
major
concerns?
Is
that
a
simple
mistake
you
know
in
a
single
network
configuration
for
example,
default
route
or
something
could
bring
down
an
entire
macro
network.
What
kind
of
tools
or
mechanisms
can
we
adopt
to
make
it
foolproof?
Well,
maybe
I'll
stop
there
and
see
if
you
want
to
save
that
learner
or
take
it
now,
yeah.
B
I
think
that's
a
great
point.
I
mean
you
know.
There
are
lots
of
stories
out
there
where
you
know
quote-unquote
fat
fingering
a
script
can
really
cause
a
lot
of
outages.
I
mean
we've
seen
horror
stories
of
VMware
administrators
in
kind
of
accidentally,
knocking
out
thousands
of
VMs
through
the
push
of
a
configuration
script
same
thing
can
happen
in
network
configuration,
especially
when
it
comes
to
route
information.
So
I
think
you
know
you
asked:
is
there
a
tool
or
a
mechanism?
B
You
know
I
would
say
personally
things
like
having
two
levels
of
approval
to
implement
changes
are
making
sure
that
you
have
someone
check
your
work
before
you
issue
it.
Those
kind
of
processes
are
always
good
to
have
whether
they're
manual
or
automated
I
also
think
that
there
are
some
things
that
you
know
when
you're
talking
about
scale
or
operating
a
production.
Maybe
they
shouldn't
be
automated.
You
know.
Automation
really
needs
to
make
sense
for
your
organization
so
and
how
you
apply.
It
depends
on
your
level
of
risk.
A
B
I
think
that's
a
great
way
to
approach
it,
just
in
general,
like
I,
said
before,
I
think
it's
also
good
to
have
more
than
one
person
be
able
to
validate
configuration,
changes
and
again,
I
think
it
also
depends
on
your
environment,
so
I
think
using
a
test
suite
or
you
know,
in
your
platform
being
able
to
kind
of
forecast
changes
or
even
pause
changes
before
you
implement
them
just
to
get
them.
Validated
is
a
great
step.
B
B
B
B
Ai
ops
platforms
enhance
IT
operations
through
greater
insights
by
combining
Big,
Data
learning
and
visualization,
and
they
produced
this
great
diagram
that
basically
talks
about
all
the
data
coming
in
whether
it's
historical
or
real
time,
funneling
that
through
and
correlating
it
and
making
in
applying
context
to
logs
metrics
wire
data
and
document
text
and
then
a
continuous
process
of
observation
engagement
and
acting
using
big
data
and
machine
learning
produces
historical
analysis.
It
improves
anomaly,
detection
and
improves
performance
analysis,
and
it
really
helps
with
correlation
and
contextualization.
B
Gardner
has
thought
about
this
particular
category
and
applied
it
to
network
management
and
has
come
out
with
this
with
a
recommendation,
or
at
least
a
forecast
that
says
by
2020
for
60%
of
network
performance.
Monitoring
buyers
will
require
improved,
AI,
ops
capabilities,
and
this
is
up
from
30%
in
2018.
So
we
can
see
that
this
trend
of
greater
complexity,
greater
embracing
and
an
implementation
of
machine
learning
is
driving
a
need
for
AI
ops,.
B
Now,
one
of
the
surprises
that
we
got
from
our
survey
was
when
we
asked
all
the
network
professionals
that
we
talked
to
about
whether
they're
actively
using
AI
ops
today,
while
the
majority
of
them
said
they're
prepared
for
full
and
network
automation
and
65%
machine
learning
is
important
for
network
management,
ai,
ai
ops,
adoption,
that's
in
the
early
stages,
in
fact
only
about
10%,
said
that
they
were
using
AI
ops
today.
So
this
is
something
that
we
found
very
interesting
and
we
found
about
it.
We
tried
to
figure
out.
B
You
know
why,
if
the
need
is
there,
why
is
AI
ops,
adoption
kind
of
relatively
low?
Well
and
really,
you
know
if
we
think
about
AI
ops,
it's
really
a
collection
of
capabilities
inside
a
platform
you
know,
and
we
see
that
while
that's
there
the
platform,
availability
and
the
platform.
Adoption
is
something
that
we
think
is
going
to
increase,
and
we
think
that
our
prediction
is
that
more
and
more
network
professionals
will
turn
to
AI
ops
platforms
to
be
able
to
kind
of
resolve
some
of
the
issues
that
they're
having
with
the
increasing
network,
comply.
B
So
a
quick
overview
of
Kent
ik
Kent
ik
is
AI
ops
for
network
professionals.
We
provide
revolutionary
analytics
to
make
every
Network
excellent.
We
have
an
AI
ops
platform
that
takes
information
from
context
and
labels,
geolocation
threat
data,
routing
and
business.
We
incorporate
traffic
and
metrics
using
net
flow
and
S
flow
and
SNMP
VPC
flow
logs
lots
of
data
for
coming
from
the
cloud,
and
we
provide
from
that
queries
and
insights
in
an
automated
way
and
actions
through
an
integration
through
lots
of
different
integration
with
stick
with
different
vendors
and
platforms.
B
B
So,
looking
back
at
the
survey,
let's
go
over
some
of
the
key
findings
and
some
of
our
recommendations.
So,
first
of
all,
it's
a
cloud
centric
world
again
everybody
on
this
call
knows
that,
and
we
appreciate
the
CNC
F,
giving
us
the
opportunity
today
to
talk
about
AI,
ops,
automation,
insights
in
relation
to
the
cloud.
We
know
that
running
today's
networks
is
dramatically
harder
and
everybody
out
there
is
really
having
a
hard
time
in
many
ways
trying
to
keep
up
and
also
think
do
things
like
prevent
outages
while
also
helping
the
business
succeed.
B
We
know
that
adoption
of
machine
learning
and
automation
are
both
on
the
rise,
and
we
also
know
that
insights
and
automation
restore
manageability
and
with
that
Ken
tech
powers,
up
network
teams
with
AI
ops,
and
we
invite
everyone
here
to
download
our
report.
You
can
get
the
link
we'll
put
it
in
the
webinar
it'll
also
be
in
the
presentation.
That's
posted
on
the
CNC
F
website
to
take
a
look
at
more
detail
and
you
can
get
a
lot
more
interesting
insights
from
the
report.
B
A
B
A
B
B
A
B
I
think
the
answer
is
that,
yes,
you
can
absolutely
have
a
demo
of
this
tool,
will
capture
your
information
and
what
it
will
reach
out
to
you
to
set
something
up,
and
in
our
demo
we
have
the
ability
to
incorporate
data
from
your
network,
your
own
routers,
and
then
allow
you
to
interact
with
the
platform
and
see
your
data.
Hopefully
that
answers
that
question
great.
A
Yeah
yeah
great
presentation,
if
anything
else,
pops
up
in
the
next
30
seconds,
we'll
catch
it
otherwise
I
know
Taylor
and
the
CNC
F
will
hand
will
get
this
getting
this
recording
online,
along
with
the
slides
which,
as
Andy
just
said,
has
some
links
there
at
the
end
for
you
to
follow
up.
If
you
want
more
info
so
yeah
thanks
so
much
everyone
for
joining
today's
webinar
and
have
a
great
day
yeah.