►
Description
In this lightboard session, Steve Tegeler analyzes the layers in an IT organization in the on-prem model starting at the datacenter and working up to developer abstractions, and explains the clear contracts between each of the layers.
A
So
what
I
wanted
to
do
is
analyze
an
on-prem
situation
and
what
that
looks
like
and
really
take,
really
go
through
some
examples
here,
all
the
way,
starting
at
the
just
the
data
center
and
working
our
way
up,
and
so,
if
we
take
a
look
at
the
data
center
here,
you
know
those
the
facilities.
Folks,
they're,
you
know
they're
they're,
choosing
various
like
a
PC
and
Leibert,
for
you
know
the
cooling
and
the
racks
and
all
sorts
of
things
and
the
clear
contract
that
they
have
with
the
physical
infrastructure.
A
Meanwhile,
the
infrastructure
team
gets
to
go
and
well
I'm
gonna
choose
a
Dell
R
610
for
my
server
and
I'm
gonna
choose,
let's
see:
EMC
/
Dell
for
my
storage,
I'm
gonna
choose
Cisco
for
my
network
and
maybe
Palo
Alto
for
my
firewall
here
and
so
provided
all
those
things
fit
in
a
19-inch
rack
and
you've
got
the
power
and
cooling
requirements
you're
good
to
go
right.
So,
let's
move
up
to
virtual
infrastructure,
so
virtual
infrastructure.
What
is
there
clear
contract
with
the
physical
infrastructure
folks?
A
Well,
for
them
it
has
to
be
x86
and
probably
on.
If
we're
doing
VMware,
it
needs
to
be
on
some
sort
of
hardware
qualification
matrix
to
ensure
compatibility.
You
need
I,
P
connectivity
and
you're,
probably
going
to
need
some
performance
from
storage
like
I,
ops,
so
from
a
performance
perspective
or
latency
to
an
even
connectivity
right.
Do
I
need
shared
storage
or
what
not.
So
that's
that's
the
the
type
of
connectivity
you
need,
or
you
may
specify,
and
that's
my
clear
contract
so
with
that
then
I
can
go
in
and
I
can
choose.
A
What
if
I
wanted
to
change
out
my
server
to
the
next
next
Dell
server
here,
like
the
dell
6600
all
right
just
made
that
up
the
dell
6600
here
has
eight
sockets
and
192
gigs
or
terabytes
of
ram
right.
It's
great
I
want
to
use
it.
I'm
able
to
change
to
that
server
because
I
haven't
changed
these
parameters,
the
x86,
the
IP
connectivity
and
as
long
as
it's
on
the
vSphere
qualification,
metrics
you're
good
to
go.
A
You
know,
maybe
I
want
to
swap
out
Cisco
for
Arista
for
switching,
because
the
wrist
is
going
to
provide
the
same
type
of
IP
IP
connectivity
to
support
nsx
I'm
good
to
go.
Ok
next
piece
is
the
developer
abstractions,
and
this
is
the
important
one
because
now
we're
talking
about
what
the
end
consumer
wants,
and
so
in
this
case
the
developer
abstractions.
Let's
call
these
out,
maybe
it's
infrastructure-as-a-service,
maybe
it's
docker,
maybe
it's
kubernetes,
or
maybe
it's
Cloud
Foundry.
A
So
this
can
be
fairly
unique
depending
on
the
technology
and
the
and
the
capabilities.
But
you
know
certainly
some
of
the
you
know
the
very
common
ones
like
take
cloud
foundry
or
kubernetes.
In
the
case
of
you
know,
we've
got
our
PKS
right,
those
all
leverage
or
those
to
leverage
a
technology
called
Bosh,
which
has
the
ability
to
deploy
down
to
the
vmware
software-defined
data
center
or
leverage
nsx
and
so
forth.
So,
finally,
here
these
are
the
developer,
abstractions
and
so
now
I'm
starting
to
provide
this
IT
service
to
the
software
developer.
A
So
if
we
specifically
call
out
some
of
the
container
ones
like
kubernetes
right,
I
want
to
be
able
to
provide
kubernetes
in
a
native
way.
I
want
to
use
upstream
kubernetes
so
that
the
software
developer,
when
they're
figuring
out
you
know
how
to
actually
architect
they
can
go
to
Google
and
they
can
go
search
and
figure
out
how
to
do
things
and
I'm,
not
using
some
proprietary
version
of
kubernetes,
which
requires
me
to
use
some
completely
different
API
and
it's
difficult
for
them.
A
So
my
clear
contract
here
as
a
software
developer
or
with
the
software
development
team,
if
I'm
running
the
developer
abstractions
layer,
is
that
I
provide
that
API
experience
and
that
could
potentially
give
me
the
opportunity
to
you
know
displace,
maybe
and
I,
as
maybe
we're
talking
about
OpenStack
right
and
maybe
I'm
running
us.
You
know
some
version
of
opus
AK,
maybe
I
want
to
provide
VMware
integrated
OpenStack,
provided
that
gives
the
same
upstream
experience
we're
good
to
go
same
with
kubernetes
and
PKS.