►
From YouTube: IETF103-BIER-20181108-1120
Description
BIER meeting session at IETF103
2018/11/08 1120
https://datatracker.ietf.org/meeting/103/proceedings/
C
B
Alright,
welcome
second
day
of
the
beer
group.
I
want
to
point
out
that
we
asked
for
two
and
a
half
hours.
They
gave
us
two
hours
and
a
another
hour
slot,
that's
a
total
of
three
and
somehow
we
managed
to
fill
it.
So
there
are
some
open
issues.
I
do
want
to
talk
about
things
come
up
here
there.
Any
discussion
worked
pretty
well
yesterday.
We
have
four
presentations
today,
be
quick
about
them.
If
you
can
and
leave
the
room
for
a
lot
of
questions.
Cuz
discussion
is
the
big
reason.
B
B
D
D
D
The
listed
requirements
must
be
supported
with
any
type
of
transport
layer
over
which
beer
layer
can
be
realized,
realized
the
repeal
protection
type
may
be
divined
and
they're
configured
from
a
centralized
controller
or
management
network,
including
peer
end-to-end
protection
and
the
link
or
load
protection,
and
it
is
required
to
support
the
failure,
detection
and
vacation
Micheli
dance.
And
finally,
it
is
required
to
support
the
fast
protection
switching
for
the
beer
package
within
the
limited
time,
and
so
we
will
add
more
requirements
in
the
future
work
if
it's
it
repaired.
D
We
provide
these
three
use
cases
for
a
beer
with
genius
under
use
case.
One
is
end-to-end
a
one
plus
one
protection.
In
this
scenario
the
multi-part
occurs.
The
traffic
must
be
sent
across
the
network
through
the
chill
disjoined
part
pathways
and
the
B
F
ers
lady,
to
receive
the
flows
transiting
from
one
of
the
toughest,
for
example,
as
the
figures
show,
there
are
two
pathways
from
source
to
destination
under
the
past,
one
is
from
the
F
IR
A
B,
C,
f
ir
v
of
ER
e
PR
PR
1,
+,
BF,
ER
and
the
past.
D
D
For
example,
the
protector
pure
link
is
from
PF
ir,
2
p
FR
a
and
the
backup
link
is
breed
pre-established
at
link
from
PF.
I
r2
p
FR
a
for
example
from
the
PF
I
RQ
p,
FR
f
and
to
the
FIA,
and
so
we
listed.
We
listed
all
the
reunion
solutions
for
beer,
including
the
use
cases,
protection
mechanisms,
detection
methods
and
the
failure
processing,
for
example,
end
to
end
1+1
protection.
The
protection
mechanism
is
send
is
to
send
traffic
across
two
disjoint
and
to
end
pathways
randolph
when
the
vania
a
creator.
D
Turner
order,
T,
IL,
f
a
Messrs.
The
details
of
these
matters
will
be
provided
in
the
future
discussion
and
and
when
their
a
veiner
agreed
or
third
load
protection.
The
traffic
away
off
switch
on
to
the
backup
link
so
for
for
this
draft,
the
Lester
step
is
to
to
to
further
research.
It
directions
like
more
resilience,
use
cases
and,
and
there
will
relate
to
them,
detection
measures
and
so
comments,
and
their
discussion
are
very
welcome.
Thank
you.
C
F
I
I
B
So
I
imagine
the
intended
adoption
who's
read
the
doc
who
feels
this
work
should
be
adopted
by
the
working
group
were
half
and
a
half
I
didn't
give.
Oh,
you
want
a
hand,
count
who
read
the
draft
hand
count
again,
three
four
five:
six:
seven,
eight
nine
nine
and
adopt
two
three
four
or
five
all
right.
Thank
you
very
much.
Thank
you
did
I
will
take
that
to
the
list.
Actually,
you
know
followed
up
in
that
comment.
Be
great
I
understand.
B
E
So
this
is
the
European
fifty
I'm,
far
away
from
the
city,
so
this
is
the
ginger
from
version
zero
and
so
on.
So
we
may
remove
that
definition.
A
war
hero,
f,
om
packet
format
and
refer
it
from
European,
because
the
bureau
am
packet.
Format
has
been
defined
in
european
so
which
has
only
refrig
and
defend
Europe.
We
have
a
tea
affair.
E
J
E
So
we
have
PR,
BFG
is
based
on
the
FDA
money
point
and
we
used
the
model,
so
there
is
no
great
way
in
and
shake
so.
So
there
is
a
comment
from
hammers
Justin
now
and
we
have
the
my
pointer
only
defined.
You
need
the
relation
and
you
unity,
unity,
unity,
relation
from
the
haunted
hotel,
so
I
think
that's
over.
E
The
unicast
should
appear
created
as
a
unique
direction,
so
I
just
I
saw
tomato
in
Manistee,
so
the
FAI
use
source
address
and
my
distributor
to
the
mat
hip
next
BFG
session,
so
so
so
source
of
the
grass
for
the
MPS
networker
wincing
intelligence.
Also,
the
gratitude
of
the
the
FI
RIT
and
vom
has
neighbor
for
the
non
ambience
network.
As
soon
as
they
saw
the
source
of
the
dress
is
the
fi
are
ident
pif
GID,
so
next
Arace
and
so
he'll,
it's
a
European
de
Kamp's
nation,
the
yeah
beer.
E
You
have
the
encapsulation
uses
a
your
om
packet
format,
the
defender
in
your
PIN
so
I'm.
The
message
that
have
a
father
must
have
you
said,
it'll
view
of
the
Appetit.
So
this,
when
you
wear
a
PTT
fine,
we
have
you
assigned
a
buyer
Ana
and
the
PFD
control
packet,
imaging
genuine
of
the
bureau.
Had,
oh
so
big.
K
When
I
saw
a
question
about
this,
so
I
mean
yeah.
Logically,
it
makes
sense
that
you
have
the
OEM
header
before
the
DFT
header,
because
since
PFD
has
to
be,
you
know,
processed
by
line
cards
and
so
on
and
should
be
as
efficient
as
possible.
I
wonder
if
it'd
be
better
to
not
use
the
OEM
header.
The
only
thing
I
am
Heather
gives
you.
It
seems
to
be
like
there's
an
OEM
type
and
some
experimental
I
mean
some
reserved
bits,
but
it
doesn't
really
add
any
value.
I
I
We
propose
to
use
what's
known
as
ACH
BFD
control
message
format,
so
without
IP
UDP
overhead,
and
thus
that's
why
there
is
the
reference
to
use
b,
FR
v,
fi
r
ID
as
identifier
of
the
source.
So
it's
a
trade-off.
Yes,
one
can
use,
for
example,
is
as
IP
packet
and
follow,
for
example,
58
84
recommendation.
So
to
use
this
donation.
L
I
H
H
H
L
I
K
Yeah,
okay,
I
agreed
it's
very
minor,
but
basically
with
this
proposal
you
you
have,
the
beer
next
header
says
is
OAM.
Then
you
have
four
bytes
and
then
you
have
an
onion,
those
four
bytes.
It
says
it's
BFD
packet
and
then
you
have
the
BFD
header
and
I.
Don't
see
why
you
need
those
four
bytes
in
its
Finnair,
because
in
the
beer
header
you
could
say
that
next
protocol
is
PFD.
I
Yes,
you
can
have
exactly
yes
there,
there
not
one
way
to
slice
the
bread
we
can
have
then
multiple
points
saying
okay
next
protocol
is
BFD.
Next
protocol
is
echo
request.
Next
protocol
is
echo
reply,
it's
just
you
know
where
you
do
the
multiplexing,
whether
you
do
it
demultiplexing
on
a
protocol
field
or
you
do
the
multiplexing
on
OAM.
H
F
I
Have
okay
in
in
the
tunnels
to
encapsulations
use,
so
it's
either
IC
use
DP,
encapsulation
or
it's
ACH.
Ach
doesn't
have
header.
So
thus
you
have
only
the
D
control
message
which
needs
to
be
identified
because
otherwise
you
don't
know
what
your
payload
is.
Okay,
so
what
stick
proposes
is
well,
it's
a
valid
approach
to
have
it
in
the
next
protocol.
But
there
is
a
concern
that
we
will
run
out
of
names
fades.
F
H
F
H
H
H
E
I
Jing
to
bootstrap
on
the
multicast
is
that
you
might
be
too
late,
so
you
either
have
to
do
it
theoretically
or
not,
and
I
can
refer
to
very
interesting
proposal
from
a
VPN.
It's
a
draft
EVP
and
fast
failover,
and
it
will
be
going
into
the
working
group
last
call
after
the
meeting
at
some
time
they
use
BGP
extension.
So
the
BGP
helps
you
to
bootstrap
point-to-multipoint
beef
this
session.
So
it
will
be
interesting
to
look
when
and
we'll
look
at
control
plane
solution
for
deer.
So
if
we
can
do
the
bootstrapping
of
this
session.
I
E
E
B
N
All
right,
so
this,
hopefully,
is
a
discussion
where
people
come
to
the
mic.
Pretty
quick,
I
hijacked
the
the
presentation
just
a
little
bit
last
night,
jim
wrong,
presented
this
last
ITF
involving
beer,
b6
encapsulation
involving
the
e6
destination
option.
Extension
extension
headers,
there's
a
variety
of
ways
to
go
about
doing
beer,
v6
encapsulation.
If
we
decide
that
that's
the
right
way
to
do
it.
N
I
know
ice
has
a
draft
out
there
that
has
been
presented
in
the
past,
involving
you
know,
putting
the
beer
bit
string
into
the
destination,
address
the
low
order,
bits
in
the
destination
address
and
high
order
bits
for
the
unicast
routing
stuff,
and
so,
and
we
look
back
at
the
the
mailing
list
to
see
some
of
the
discussion.
That's
happened
with
beer
v6,
and
maybe
you
can
set
me
straight
on
this.
N
H
B
We
create
an
abstraction
that
the
address,
but
now
we're
putting
reference
into
the
address
and
we're
looking
at
hardware
spin
anyway
to
get
that
kind
of
support
and
the
hardware
spin
for,
for
extension,
headers
and
v6
would
be
a
requirement
as
well
so
I'm
trying
to
understand.
What's
unique
about
that
use
case,
when
we
having
specific
beer
header,
we
had
the
encapsulation
discussion
and
the
off-site
interim
working
group
meeting
where
we
tried
to
when
we
did.
B
We
accomplished
a
beer
header,
consistency
between
native
and
cap
and
Ethernet
ready,
she's
me
mpls
in
camp,
and
we
had
cohesion.
We
move
forward,
and
so
the
v6
hop
aha
fit
into
that
space.
We
didn't
modify
it
I'm,
not
saying
we
don't
need
to
find
a
way
to
do
it,
but
I
think
we
at
work
at
a
maturity
point
if
I
dare
say
maturity
where
we
need
to
better
understand
why
this
is
different.
B
G
Who
amber
tolling,
Nokia,
yeah
I
think
I?
Second,
that
the
beat
of
forwarding
is
done
via
the
beer
index
or
the
BET's,
and
we
need
something
to
identify
that
this
is
a
beer
header,
so
I
don't
see
why
that
identification
has
to
be
different.
Ip
v4
ipv6
like
to
me
it's
it's
the
same
thing
in
both
atmosphere.
N
Agreed
and
we've
discussed
that
previously
so
I
guess
the
question
and
again
I
there's,
maybe
not
a
big
need
to
go
into
details
of
this
particular
in
cap
proposal,
because
there's
been
a
few
of
them,
but
is
there
a
need,
based
upon
your
and
Greg's
comments,
to
have
a
problem
statement
trap
that
discusses
the
use
cases
of
whether
we?
This
is
something
that
we
need
to
pursue
within
this
written
group?
Is
that
I?
Don't
we
don't
want
to
do
it
if
it's
not
something,
that's
necessary
sure
Gregor's.
B
H
Okay,
so,
for
example,
the
HomeNet
is
simply
to
the
chips.
Lots
of
chips
cannot
take
a
new
mac
type
easily
and
they
surely
will
not
support
for
quite
a
time
right.
So
the
only
solution
was
going
link-local
in
local
very
consciously.
So
you
cannot
easily
go
multiple
hops
right,
but
if
this
has
a
use
case,
section
that
shows
cannot
do
that.
Fine,
it
doesn't
well.
We
have
a
bunch
of
proposal
on
the
table
right,
so
you
can
click.
N
O
O
That
also
means
that
the
let's
say
the
process
that
dealing
with
the
the
3/6
forwarding
can
a
look,
has
access
to
the
bits
itself
and
you
can
do
be
specific
processing
in
you
inside
yuffie
six
code
right
so,
and
that
made
it
so,
let's
say
we
did
a
hackathon
at
the
time
right.
I
think
the
first
hackathon
that
was
done
in
IDF
ira
prototype
that
onto
a
linux
kernel.
O
So
there
was
quite
easy
to
because
you
have
direct
access
to
those
bits
because
they're
in
the
the
so
that
made
it
easier
to
get
some
beer
falling
going
and
not
have
like
a
generic
beer
forwarding
code
somewhere
sitting
in
the
kernel.
So
I
think
that
was
one
of
the
attractions
we
had
so
so
haven't
done
much
work
recently
on
this.
So
we're
not
really
sure
yet
what
to
do
with
draft
and
it's
all,
depending
on
use
cases
right
and
right,
no
problems,
wanna
solve
so
I.
O
N
B
B
It's
when
we
see
this
and
I've
had
talked
to
Cisco
engineers,
who
were
v6
zealots
who
were
took.
You
know
two
hours
of
wine
and
beer
for
them
to
realize.
If
you're
doing
this
to
be
six,
it's
not
be
six
and
their
whole
hand
was
oh,
let's
put
this
and
be
six
so
I'm
gonna
put
you
on
a
pedestal
or
give
beer
a
leg
up
with
v6
Connect
or
something,
but
to
my
point
once
you
do
that
it's
not
v6
anymore
you're,
just
kind
of
misappropriating
the
bits
of
some
way
but
I.
B
Think
in
in
the
use
case
for
the
draft
that
that
pier
put
together
the
hybrid
or
bid
still
had
a
v6
purpose.
It
wasn't
like
just
taking
entire
group
and
the
idea
I
mean
it
was
questionable
whether
there
was
security
issues
with
the
six
six
and
whether
that
use
case
made
sense,
but
he
at
least
to
find
a
use
case
where
they
were
forwarding
on
the
high
order
bits.
So
they
got
to
the
beer
domain
and
then
blew
up
the
bits
down
below
on
his
beer.
Okay,.
B
N
Next,
anyway,
no
I'm
not
wrong
is,
but
so
we
we
encourage
you
to
look
at
these
other
slides
there's.
We
do
have
a
slide
on
like
kind
of
recapping,
some
of
the
discussion
that
has
happened
on
the
list
eric
rose
and
had
a
lot
of
good
opinions
on.
If
we
are
to
to
do
something
like
a
beer
v6
in
cap,
then
you
know
these
are
some
things
we
should
consider.
So
we've
got
that
and
we'll
include
that
in
our
drive
stigman.
K
On
so
I
have
so
many
a
specific
comments
back
and
give
you
those
later
but
I
think
what
could
be
interesting,
what
be
to
have
an
ipv4,
ipv6
and
cap
so
that
you
can
send
a
beer
packet
through.
You
know
IP
only
routers,
basically
bypass
non-beer
robbers,
but
that's
not
what
this
draft
is
doing.
That's
a
different
thing:
I
guess.
N
N
B
M
P
The
draft
had
updated,
so
it's
six
six
version
main
part
of
this
draft
include
sometime
important
terms:
P
Italo
hit
remit
hollow
and
PGP,
and
maybe
in
FEC
and
one
more
one
more
time,
pseudo
PRF
on
segmented,
appoint.
This
is
for
IP
lookup,
so
Tobias
is
comparable
to
the
PDP
and
maybe
a
few
see
represented
by
the
route
IP
and
the
arty.
P
The
same
concept
is
many
ppl
hellos
can
show
the
same
route
IP
NRT
and
will
be
mapped
to
the
same
suit
over
F
and
generic
concept.
Its
tallow
sticking
can
be
between
any
two
of
the
amount
if
he
sweetie
or
IR
or
peer
Paulo,
and
that
the
under
to
indistict
tallow
can
be
bound
to
one
or
many
PDP
and
maybe
a
VC
from
some
ingress.
He
and
the
graph
and
the
university
can
decided
to
use
which
tallow
for
which
flows.
P
P
These
comments
came
from
Eric
when
I
discussed
with
the
authors
of
BM
a
beam
before
this
zero
draft
and
I
found
that
the
IP
no
cap
exists
in
the
0
3
3
version
of
this
draft
and
the
be
removal
was
lost
in
the
4th
first
getting
drafted
and
also
the
the
pika
probe.
The
main
problem
we
want
to
solve
the
that
multicast
join
latency
was
added
as
more
efficient
in
the
maybe
a
explicit,
the
tracking
draft.
So
we
agree
with
the
text
very
much
and
I
think
it.
P
P
There
is
already
defined
RFC
GTM
using
IR
has
very
similar
option
to
use
a
positive,
join
leave
a
tea
from
downstream
and
the
TEM
using
IR
has
a
very
similar
option
to
use
IP
lookup
when
aggregating
flows
using
one
label.
So
this
I
think
a
dystrophy
is
very
similar
to
the
proposal
of
positive
joint
initiated
by
our
RPF
and
allow
the
aggregating
flows
using
one
label
and
the
pair
flow
replication
without
flooding
using
a
vlookup.
P
P
P
P
If
we
do
without
a
be
no
cap,
then
we
can
look
up
the
Vivian
labor
for
bit
stream
directly,
but
this
needed
allocating
of
bacon
label
for
every
say
flow
before
from
the
foot
forward
in
cycles,
I
mean
okay,
we
all
need
more
forward
in
cycles
forward
in
people
rose
and
the
way
I
window
cab
will
need
T,
plus
and
states.
He
represented
the
sticking
palace
and
and
represented
the
number
of
floors.
P
He
can
be
one
in
some
case
and
the
ends
days
is
or
I'm
Table
six.
Of
course
they
are
wider
than
the
median
label
and
that
he
is
possibly
can
be,
and
in
some
case
in
this,
in
such
a
case,
of
course,
the
I
hit
OK
will
cost
will
cost
cost
more
so
summer,
summary
I
begin
dough.
Cap
can
help
to
leverage
is
the
more
efficient
our
RPF,
explicit
shaking
and,
of
course
it
has.
The
cost
has
a
recording
cost.
So
this
is
the
positive
and
the
negative
of
this
of
this
proposal.
P
Q
Children
from
juniper
I
replied
on
email
list,
saying
that
all
the
discussions
still
commence
me
more,
that
this
is
not
specific
to
beer,
even
except
that
the
beer
and
making
draft
said
that
you
know
you
stays
explicit
other
the
RPF
experience
it's
pretty
tracking
when
you
do
the
segmentation
anything
else,
it's
generic.
It
applies
to
any
tunnel.
Yes,
my
first
point
and
the
second
point
so.
Q
I
love
segmentation.
There
are
scenarios
where
you
do
need
segmentation,
but
every
time
when
bring
up
segmentation
customer
says
that
means
I
need
to
maintain
States
on
the
segmentation
points
and
now
you're,
adding
even
more
you're,
adding
the
IP
lookup
the
works
under
sickness
nation
points
that
will
make
it
even
less
attractive.
P
P
Q
Q
You
in
here
you
only
care
about
beer,
but
we
have
a
more
generic
problem
to
solve
it.
We
do
not
have
to
limit
the
scope
to
this
to
beer.
Specifically,
we
can.
We
can
solve
this
problem
that
covers
any
panel
type.
Basically,
if
you
need
to
do
a
segmentation-
and
you
cannot
do
per
flow
label,
then
just
to
IP
lookup
and
whether
its
beer
tunnel
or
ingress
replication,
whatever.
P
P
P
P
Q
H
H
This
in
a
sense
is
putting
more
state
back
into
the
network
or
even
more
stating
previously
right,
so
you
still
FM
LTP
and
have
your
beer
on
top.
Of
course,
we
can
build
any
possible
combinatorial
of
all
the
multicast
technologies
with
beer,
but
I.
Don't
think
this
is
something
this
group
should
you
know
based
on
the
Charter
and
the
original
direction
attempt
to
do
so.
H
B
B
B
K
B
G
G
K
G
Maybe
in
each
other's
discussion
about
this,
but
the
beer
is
where
the
OIF
comes
from
and
it
builds
a
table
based
on
that,
not
anything
to
do
with
pim
itself,
so
you
can
note
it
like
any
other
protocols
out
there
and
send
it
to
p.m.
via
whatever
structure
you
want
to
use.
But
again
to
me
this
is
implementation.
This
is
not
protocol.
I.
O
Cisco
so
remember
that
in
beer
there's
also
subdomains.
So,
although
you
know
to
be
a
VFR
ID,
that's
in
the
packet,
it
doesn't
tell
you
what
subdomain
of
the
forwarded
on
right
now,
implicitly
that's
encoded
into
the
amperes
label,
but
it
is
a
bit
harder
to
figure
out,
because
the
label
doesn't
tell
you
anything
right,
there's
no
context
that
comes
with
the
label.
She
has
to
look
into
beer
to
figure
that
out
so
from
a
design
point
of
view
which
it's
it's
nicer.
O
G
So
that's
fine.
We
can
add
the
IB
BR
IP
address
as
source
I'm
guessing.
But
then,
when
you
remove
the
beer
header,
the
extraction
becomes
the
issue
because
we
decided
to
send
that
as
an
IP
protocol.
Not
we
didn't
come
up
with
a
brand
new
protocol
for
it.
So
you
remove
the
beer
header,
it's
IP
protocol.
What
you
gotta
do
now.
K
More
or
less,
my
point
is
that
once
you
remove
the
beer
header,
you
lose
that
context
about
the
source,
beer,
Freddy
and
so
on
that
you
have
in
that
header.
So
when
even
gets
say
this,
it's
a
pure
IP
packet
and
and
then
he
needs
some
way
of
finding
out
what
the
source
before
idea
was.
So
the
easiest
way
is
to
put
that
in
a
joint
packet
itself.
I
should
add.
This
is
also
an
issue
for
the
IGMP
ola.
They
may
need
to
some
way
of
putting
this
in
the
rtmp
packet.
M
Q
It's
not
only
the
source
address,
but
also
sub
domain
ID
data
converting
so
in
theory,
you
can
use
the
incoming
beer
label
and
the
beer
header
to
figure
out
other
information,
but
indeed
typically
when
you,
when
you,
when
you
send
a
packet
up
the
stack
you
remove
those
headers
one
by
one
so
most
likely,
we
may
be
not
the
case
in
your
implementation,
but
most
likely
when
pin
gets
the
packet.
All
you
have
is
just
a
pin
packet.
H
The
to
box
is
your
vertical
stack:
you
carry
through
whatever
you
need
to
get
the
job
done
or
heat.
Actually,
however,
I
think
the
limitation
is,
then
that
you
bgp
and
you
pin,
join
and
your
beer
and
point
all
have
to
be
co-located
right
and
don't
we
accept
these
architectural
limitation.
I
mean
I'm,
not
making
a
suggestion
forward
backwards,
but
I
do
I
mean
it's,
surely
not
our
mandate
to
determine
how
people
implement
this
stuff
and
it's
a
viable
implementation
within
the
architectural
framework?
Oh
well,
all
that
so
loose.
G
So
again,
I
think
the
first
draft
that
we
gave
out
was
trying
to
solve
this
issue
by
coming
up
with
a
new
beer
type
packet
type.
That's
it
you
know
signaling
and
and
then
we
were
putting
the
IP
address
of
the
IB
BR
as
a
source
IP,
but
it
was
he.
It
was
said
that
that's
not
the
right
thing
to
do
so.
We
we
went
down
the
different
paths
and
like
not
putting
anybody
on
this
part.
I.
Think
Jeffrey
was
yourself.
That
said,
we
need
to
not
come
up
with
a
brand
new
protocol.
Q
No,
that's
not
a
brand
new
protocol.
Indeed
you
are
you
just
in
a
put
up
and
signaling
IP
practice.
That's
all
you
need
to
do
at
that
time.
I
attest!
That's
what
I
think
and
now
after
stick
pointed
out,
I
do
realize
that,
yes,
your
implementation
could
figure
out
that,
oh
that's,
incoming
prefix
appear
prefix
and
the
subtle
memory,
or
you
could
figure
it
out
in
your
implementation
by
picking
back
up
all
those
information,
but
a
more
easier
way
here
would
be
to
to
send
the
information
along
with
in
the
pin
packets.
Q
K
I'm
also
happy
with
sending
a
pin
joint
directly
away
beer.
The
way
it
is
now
just
would
like
to
have
that
information
inside
the
joint
but
yeah
I
would
say
when
I
wrote,
you
know
the
idmp
draft
and
implemented
that
what
I
did
was
just
use
the
source
address
of
the
MP
right,
assume,
that's
a
beer
prefix
and
you
start
to
find
a
be
Friday
and
you
can
do
with
that
with
PIM.
The
problem
is,
you
don't
know
the
subdomain,
though
so
I,
basically
just
used
CLI
to
say:
okay,
any
I
am
Pete.