►
Description
A presentation hosted by Hyperledger's Telecom Special Interest Group. In this talk, Nima Afraz, Ph.D., Postdoctoral Researcher at CONNECT Center in Trinity College Dublin discusses how Hyperledger Fabric could facilitate the implementation of a distributed marketplace for 5G network infrastructure sharing which does not rely on a central source of trust.
More about the Telecom Special Interest Group is at: https://wiki.hyperledger.org/display/TCSIG/
----
Hyperledger – Open Source Blockchain Technologies
Hyperledger is an open source community focused on developing a suite of stable frameworks, tools and libraries for enterprise-grade blockchain deployments.
It serves as a neutral home for various distributed ledger frameworks including Hyperledger Besu, Fabric, Sawtooth, Indy, as well as tools like Hyperledger Avalon, Cactus and libraries like Hyperledger Aries, Ursa. Learn more about Hyperledger projects: https://www.hyperledger.org/use
Case Studies: https://www.hyperledger.org/learn/case-studies
Training & Certification: https://www.hyperledger.org/learn/training
Tutorials: https://www.hyperledger.org/use/tutorials
Webinars: https://www.hyperledger.org/learn/webinars
Events: https://www.hyperledger.org/events
Vendor Directory: https://www.hyperledger.org/use/vendor-directory
Subscribe to the Hyperledger Newsletter: https://www.hyperledger.org/newsletter
Follow-us on Twitter @Hyperledger
Learn about Hyperledger Membership: https://www.hyperledger.org/about/join
#Hyperledger #HyperledgerTelecomSIG #Blockchain
A
A
C
A
A
A
A
Dorman,
would
you
like
to
it
sounds
like
vivin
is
having
a
problem
with
this
mike?
Would
you
like
to
maybe
introduce
not
to
put
you
on
the
spot,
but
would
you
want
to
maybe
just
explain
to
people
who
are
new
to
the
call
what
this
group
is
and
and
then
after
an
introduction
and
sending
some
context,
we
can
then
hand
it
off
to
nema.
D
Yeah
sure
yeah
sure,
thanks
david
and
yeah
good
evening
good
afternoon.
Everyone
very
excited
to
see
you
all
here
and
yeah,
I'm
I'm,
although
I'm
pretty
new
to
this
group,
but
very
excited
to
be
here.
So
as
we
all
know
this
is.
This
group
is
primarily
for
a
special
interest
group
for
telecom,
where
you
know
where
we
have
the
like-minded
and
the
experts
from
various
parts
of
the
world
who
are
interested
in
kind
of
you
know,
bringing
in
new
ideas
or
working
together
and
research.
D
What
we
can
do
in
this
telecom
sector
with
regards
to
blockchain
and
new
areas
of
interest,
including
the
5g
iot,
the
interconnect
and
all
that,
and
so
it's
a
special
group
wherein
you
know
everybody
is
invited
and
more
the
merrier.
Basically,
we
kind
of
meet
fortnightly
and
then
discuss
on
the
projects
that
are
going
on.
D
The
idea
is
to
kind
of
bring
in
newer
ideas,
put
together
some
white
papers,
research
on
the
ideas
and
then
bring
out
some
standards
in
this
area,
wherein
we
can
actually
evangelize
and
put
together
some
standards
for
better
adoption
of
blockchain
in
telecom
space
yeah.
So
that's
primarily
the
thing,
and
today
we
have
a
very
exciting
talk
from
nema,
wherein
probably
I
would
let
maybe
david
or
give
that
introduction
and
get
started
over
there.
A
Great
thank
you
for
that
darman
and
one
note
just
to
let
everybody
know
I
did
drop
a
link
in
the
zoom
chat.
If
you
are
new
to
this
group
and
are
interested
in
joining
the
discussion
beyond
this
call,
we
we
have
an
open
forum
where
you
are
welcome
to
join.
We
have
a
mailing
list
and
then
we
have
regular
calls
if
you
join
subscribe
to
that
mailing
list.
There's
a
link
in
the
chat
you're.
Welcome
to
introduce
yourself
ask
questions
of
the
group
join
in
discussions.
A
If
you're
on
the
mailing
list,
you'll
also
get
notified
about
upcoming
calls,
we
do
meet
regularly
and
we
do
have
various
guest
presentations.
So
if
you
want
to
find
out
about
those
just
feel
free
to
sign
up
to
that
list,
and
with
that
let
me
introduce
nima.
He
is
a
long-time
member
of
this
group.
He
recently
offered
to
share
some
of
the
research
that
he's
doing
so.
We
appreciate
his
time
and
today,
he's
gonna
share
about
decentralized,
5g
marketplaces
based
on
hyperledger
fabric.
So
with
that
I'll
hand
it
up
to
nema.
B
So
today,
I've
been
really
I'll,
be
talking
about
a
problem
that
is
related
to
the
research
that
we've
been
doing
in
the
research
center,
regardless
of
the
of
the
blockchain
work.
So
this
is
mainly
really
about
the
resource
allocation
problem
for
the
for
the
next
generation
of
the
communication
networks
and
mobile
communication
networks,
which
is
the
the
5g.
B
B
So
this
really
means
that
the
smaller
and
more
innovative
service
providers
want
to
provide
kind
of
very
creative
services
over
over
the
network.
They
won't
be
really
able
to
kind
of
get
the
the
resources
and
the
infrastructure
they'll
need
for
to
join
these
networks
and
serve
the
their
customer
base.
B
So
these
include
really
the
the
radio
spectrum,
the
base
stations,
the
fiber
and
all
the
equipment.
Essentially.
So
this
really
makes
it
very
important
for
us
to
enable
the
the
secondary
usage
of
5g
network
infrastructure,
meaning
that,
as
as
as
it's
been
the
tradition
in
in
telecom
networks,
the
bigger
operators
do
share
their
resources
with
others.
B
So
currently,
the
way
it's
happening
is
over
very
long
term
kind
of
contracts
between
these
network
operators.
But
what
what
we
see
is
that
those
kind
of
contracts
are
not
going
to
be
sustainable
for
the
for
the
5g
network,
so
just
some
examples
of
the
resource
sharing
the
secondary
resource
sharing
we're
talking
about
in
5g.
So
this
could
be
radio
spectrum
sharing
the
secondary
radio
spectrum
sharing.
B
These
are
also
kind
of
networks
that
could
be
shared
within
multiple
operators
and
in
the
secondary
resource
sharing
model.
So
this
slide
is
really.
What
was
trying
to
tell
is
that
the
the
technical
infrastructure
is
out
there
and
they're
sandra
the
the
most
major
standardization
bodies,
like
fcc
fcc,
5gbp
broadband
forum.
B
B
B
So
here
you,
you
kind
of
see
how
the
scheme
of
the
the
slice
brokering
example
in
the
in
the
5g.
So
this
is
the
centralized
model
where
you
have
the
the
mvnos
and
otts
and
infrastructure
providers
in
the
left-hand
side,
so
assuming
that
they
have
some
idle
resources
and
then
you
have
on
the
other
side
on
the
right
hand,
side
at
the
bottom.
You
have
the
nvnos,
the
notps,
who
are
in
demand
of
certain
resources
that
are
going
to
be
dedicated
to
them
in
as
a
slices.
B
So
these
slices
could
include
some
run.
Computing
storage
could
be
a
combination
of
each
and
in
the
middle
you
have
the
slicing
manager
or
orchestrator
and
the
slice
broker.
That
is
a
kind
of
a
part
of
that
management
system
where
in
in
in
our
case,
we
are
using
auctions
and
as
a
market
mechanism
to
really
allocate
those
idle
resources
to
the
people
who
are
kind
of
bidding
for
them.
B
We'll
we'll
talk
about
that
a
little
bit
later
on,
so
all
good.
So
far,
let's
say
previously
conventionally
you
would
have
a
centralized
kind
of
infrastructure
provider,
maybe
a
government
entity
or
maybe
a
major
operator
that
all
the
smaller
operators
trusted
and
then
they
would
usually
take
care
of
this
market
negotiation
and
market
mechanism,
and
they
would
run
this
market
mechanism
and
then
everyone
would
be
happy.
Everyone
would
be
dedicated
their
their
resources
and
and
all
is
good.
B
B
B
B
B
So
essentially,
we
are
seeing
the
the
network
slice
as
a
commodity
here
and
for
this
for
purpose
of
this
research.
For
now
we
are
defining
the
the
slices
kind
of
as
in
templates,
and
then
the
the
the
users
can
buy
multiples
of
that
particular
slice
unit
again,
so
another
really
way
of
showing
how
how
the
how
the
system
works
so
really
for
our
reasons
for
for
choosing
fabric
over
over
other
available
platforms
was
the
model
or
design.
B
So
as
as
you
know,
latency
is
really
the
the
the
trickiest
part
in
5g,
and
the
numbers
are
really
very
strict
and
also
the
multi-channel
support.
Again,
that's
that's
very
beneficial
when
you
say
you
have
20
network
operators
who
want
to
do
this
trade
with
each
other,
but
they
don't
necessarily
want
to
reveal
all
the
data
or
all
the
information
with
all
the
other
operators,
so
they
can
just
create
their
own
channels
and
kind
of
talk
on
that
channel
with
a
smaller
group
of
operators
again.
So
these
are
really
the
transaction
flow.
B
So
this
is
how
our
deployment
works.
So
we
use
a
public
cloud
service,
so
we
really
just
deploy
the
components
on
different
vms
that
represent
different
organizations.
We
use
docker
swarm
to
orchestrate
the
the
the
kind
of
harborage
of
fabric
dockers.
Again,
we've
used
some
other
technologies
to
do
the
benchmarking
we
use
hyperlogic
caliper
to
benchmark,
and
then
we
use
grafana
proteus
and
some
other
tools
for
for
monitoring
of
these
kind
of
use
case
that
we
have
here.
B
B
B
What
we
are
currently
working
on,
it's
taking
a
lot
of
our
time
here
is
really
trying
to
so.
As
I
mentioned
with
the
slicing
use
case,
we
are
already
meeting
the
required
demand,
but
for
for
having
this
marketplace
distributed
marketplace
platform
to
be
used
for
many
other
use
cases,
we
could
try
to
see
how
much
it
scales
really.
So
this
is
really
the
numbers
that
we
are.
We
are
achieving
at
the
moment,
so
the
suts
are
just
the
systems
under
test.
B
So,
as
far
as
I
remember,
the
the
sut-1
is
is
two
organizations
on
on
three
vms.
So
two
vms
are
hosting
the
the
organizations.
They're
they're
reached
32
cores
and
similar
ram,
and
then
you
go
forward
and
then
you
have
five
organizations
again
with
different
hardware,
and
then
you
have
the
transaction
throughputs
that
we've
been
able
to
achieve
up
to
400
at
the
moment,
with
the
current
auction
algorithm
and
still
kind
of
the
the
latency
lower
than
one
second.
B
B
So
here
we
didn't
really
worry
about
the
interference
with
other
networks
and
so
on
or
inter
slice,
interference
and
so
on.
So
no
so
this
is
not
really
a
part
of
part
of
this
three
stage,
but
we
are
at
the
moment
trying
to
find
the
right
kind
of
sdn
controller
for
for
these
grand
networks
that
are
already
enabling
network
slicing
and
we're
trying
to
integrate
the
solution
with
those.
When
we
have
really
future
plan
to
extend
this
work.
F
B
So
what
we
are
here
doing
is
that,
although
that
the
brokering
is
already
automated,
but
it's
automated
with
the
assumption
that
you
have
that
central
intermediary
who
takes
care,
takes
care
of
execution
of
these
kind
of
whatever
market
mechanism
is
there?
So
if
you,
if
you
go
online
and
search
for
for
these
slice,
brokering
or
spectrum
secondary
market
kind
of
sharing,
you
will
see
plenty
of
algorithms
that
are
used
to
kind
of
facilitate
this
trade.
B
B
So
this
won't
be
really
a
very
small
kind
of
network
function
that
sits
somewhere
in
the
controller.
So
you
will
need
a
kind
of
a
decent
enough
kind
of
infrastructure
computing
power
to
be
able
to
run
these
right.
But
then,
if,
if
that's
not
really
the
case,
you
can
actually
have,
for
example,
the
the
the
the
xilinx
boards
that
the
previous
weeks
they've
been
talking
about
them
here
here
in
the
and
the
telecom
stick.
F
Yeah
this
one,
so
so
so
this
suv
one
is
like
you
are
using.
Let's
suppose,
two
operators,
as
you
do
like
two
you
are
using
different
operator,
means
in
if
you,
if
I
assume
it
like,
let's
say
different
two
vms
are
in
sc2.
We
have
three
or
four
vms
and
then
we
are
trying
to
see.
I
mean
what
is
this:
is
it
latency
throughputs
and
what.
B
B
I
said
it's
more
than
what
we
need,
so
that
could
really
meet
the
requirements
of
the
5g
network
slicing,
but
then
we're
looking
at
other
use
cases
that
could
use
this
particular
distributed
market
model
or
the
marketplace
that
we
are
kind
of.
So
that's
really
the
main
product
product
that
we
have
here
right.
B
B
So
what
that
means
is
that
they
are
there.
Are
I
mean
what
we
initially
designed
this
auction
algorithm
was
for
those
particular
networks.
Then
we
wanted
to
share
them
with
with
multiple
operators.
So,
obviously
back
then
we
weren't
really
working
on
blockchain.
We
were
just
trying
to
see
how
these
kind
of
how
we
can
actually
design
an
algorithm
that
would
respond
to
such
a
high
frequency
kind
of
algorithmic
market
mechanism.
B
So
here
so
again,
so
if
you
want
to
kind
of
consider
such
scenarios
for
those
numbers,
so
you
imagine
125
mark
second
per
millis
transactions,
no
transactions,
every
100
by
25
microseconds.
That
means
8
000
transactions
per
second
and
okay.
So
I
mean
there
there
are
kind
of
some
papers
and
some
work
that
shows,
let's
say:
there's
there's
fast
fabric.
They
claim
that
fabric
can
support
up
to
20
000
transactions
per
second.
Well,
obviously,
that's
not
the
standard
type
of
ledger
fabric,
so
they
they.
They
do
lots
of
improvements
there.
B
So
they
take
out
all
the
nodes
from
the
docker
that
docker
containers
they
put
them
straight
on
the
the
system
itself.
They
they
store
that
the
the
blocks
or
the
transactions
on
the
on
the
ram.
Let's
say
instead
of
the
disk
and
many
other
improvements,
and
they
they
claim
that
they
can
achieve
those
those
numbers
so
and
they
they
showed.
They
actually
illustrate
that
so
we,
what
we
are
trying
to
achieve
here
is
to
see
what
our
kind
of
smart
contract
that
is
the
auction
can
achieve.
B
So
here
the
the
best
number
so-
and
this
figure
is
really
I
just
generated,
maybe
an
hour
ago.
So
this
is
really
the
latest
results
that
we
are
getting,
so
we
haven't
been
able
to
reach
more
than
400
transactions
per
second-
and
that's
that's
quite
a
big
number,
if
you
kind
of
consider
it
in
business
terms,
but
it's
still.
If,
if
it's
going
to
be
used
for
such
scenarios,
that
I
mentioned
there
is
still
some
some
way
to
go
and
really
so.
F
Yeah
thanks
thanks
nima.
I
think
I
also
worked
with
this
5gpp
kind
of
things
and
we
worked
on
network
slicing
and
yeah.
That
is
like
a
very
good
useful
because,
because
I
worked
on
end-to-end
network
slices
and
and
these
slices
with
sdn
controllers,
we
had
one
orchestrator
and
we
use
some
kind
of
orchestra.
We
used
blockchain
kind
of
orchestrator,
I
think,
maybe
next
to
next
time.
I
can
show
that
yeah
we
can
discuss
over
this,
but
we
have
more
questions
here.
Brett
is
asking
sqldb.
I
think
we,
it
is.
B
F
Yeah,
even
even
every
hyperledger,
even
every
blockchain,
most
of
the
blockchains
are
using
like
levelly
because
db
things
so
yeah.
So
we
have
one
question
from
mazda
he's
asking
with
centralized
property
entity,
actual
sharing
of
slices
result
or
resources
is
guaranteed.
Can
we
have
a
guarantee
mechanism
here.
B
Yeah,
so
that's
a
that's
a
great
question
mazda.
That
is
actually
really
the
next
step
that
we
are
taking
towards
towards
this
work
so
and
we
do
actually
have
a
sla
sub
group
in
in
hyperledger
telecom,
special
interest
group,
and
it's
been
it's
been
idle
for
a
while,
but
we
actually
encourage
people
to
join
and
contribute
work
there.
So,
yes,
so
essentially
the
smart
contract
that
we
have
here
is
the
auction
and
market
mechanism
smart
contract.
But
this
smart
contract
could
be
easily
extended
to
kind
of
monitor
way
more
features.
B
So
you
can
essentially,
let's
say,
define,
availability
there
and
let's
say
you
want
99.9
availability
for
that
for
the
network
slice
that
you
have
acquired
and
then
you
can
have
certain
measurement
points
set
up
on
your
on
your
network
and
those
values
could
be
recorded
on
the
ledger
and
then
the
smart
contract
could
actually
enforce
those
requirements
that
are
essentially
in
an
if
and
else
so.
Smart
contact
could
be
just
if
and
out.
B
B
B
B
So,
essentially,
a
slice
is
1x
run
one
x
computing,
one
exit
storage,
so
that's
the
template.
So
this
is
one
unit
of
a
slice
and
then
this
one
unit
of
a
slice
is
the
commodity
that
we
are
trading
here
in
the
in
the
market,
so
let's
say
mvno
one
bids
for
30,
sl,
30
units
of
network
slice
and
that
and
then
with
a
certain
kind
of
amount
of
money
that
he's
willing
or
she's
willing
to
pay
for
this
particular
amount
of
sliced
resources.
B
That
being
said,
this
is
the
the
most
kind
of
simplified
scenario
that
we
have.
If
you
want
more
sophisticated
market
mechanism,
they
are
available.
So
the
the
auction
mechanism,
I'll
be
using
here
is
as
a
double
auction,
is
a
multi-item
double
auction.
But
you
can
go
further.
You
can
use
combinatorial
auctions
that
actually
allow
kind
of
gives
you
more
freedom
to
actually,
let's
say,
source
this
slice
from
multiple
operators.
So
you
can
get
your
run
from
one
operator.
B
B
A
B
Again,
so
I
I'm
not
really
sure
how
this
would
translate
into
actual
slices
in
the
controller
right.
I
just
I
just
know
that
it
would
be
translated,
because
these
systems
are
already
working
just
just
without
the
blockchain
right,
so
yeah
the
infrastructure
is
there,
so
the
technology
is
there
so
yeah
I
mean
I
I
I
assume
using
the
rest
api.
B
Probably
you
can
actually
transfer
this
information
or
that
or
the
controller
could
read
this
information
as
queries
to
the
to
the
to
the
state
database
and
then
when
we
could
read
that,
let's
say
in
this
period
of
time
operator
p
has
gotten
this
amount
of
resources
from
operator
x,
and
then
they
can
do
that.
The
allocation
part
there.
C
B
If
we
have
10
operators
in
the
in
the
network,
I
would
understand
having
10
channels
right,
so
one
one
between
each
operator,
let's
say
or
well,
that's
actually
more
but
yeah
anyway.
So
I
can't
I
can
imagine
I
can
understand
that.
But
I
don't
really
see
a
point
because
initiating
this
channel
on
the
on
the
ledger
it
takes
takes
takes
some
time
and
it
takes
configuration
and
so
on,
and
I
I
don't.
I
don't
really
see
why
we
would
need
a
channel
per
personalize.
E
E
E
B
C
C
I
I've
ever
been
hired
nemo
so
yeah.
I
I
got
a
couple
of
questions.
So
first
thing
is
related
to
this
performance
figures,
which
is
really
interesting.
Probably
what's
what
I
see
may
be
missing-
or
maybe
I
didn't
catch-
is
that
how
different
other
configurations,
let's
say,
endorsement
policy
number
of
channels
etc,
would
affect
this
or
like
bring
the
numbers
up
or
down,
because
with
endorsement
policies
that
would
require
more
validations
from
different
peers.
B
B
B
We
really
want
every
single
operator
to
execute
this
auction
and
we
want
all
of
their
results
to
be
the
same,
so
that
that's
why
we
actually
chose
that
yeah
in
our
event
and
regarding
the
channels
for
now,
we
only
have
one
channel
because
we
are
not
really
implementing
the
private
data
part.
So
that's
that's
not
really.
Our
concern
at
the
moment.
B
B
I
have
been
experimenting
with
from
10
to
400
really
and
it's
been
actually
frustrating,
because
I've
been
trying
to
kind
of
play
around
with
the
block
sizes
yeah
and
achieve
higher
transaction
throughputs,
but
I
haven't
been
able
to
actually
so
450,
for
now
at
least
has
been
really
the
highest
throughput
that
I've
been
able
to
achieve
with
reasonable
latencies
right.
So
I
mean
once
you
really
hit
the
hit
that
threshold
of
your
of
your
network
and
of
your
system.
B
So
I
I
was
going
mad
actually
and
I
was
trying
to
use
like
128
cores
per
each
kind
of
pier
and
so
on,
but
I
mean
at
some
point
those
resources
really
don't
matter
anymore
and
it
has
you
kind
of
reach
the
limitation
of
the
of
the
fabric
components
themselves.
So
again,
there's
as
I
I
know
that
you
have
a
deployment
experiment
from
from
the
previous
use
cases,
I've
been
working
on
and
and
actually
it's
it's
it's
becoming
really
difficult
to
kind
of
take.
B
C
It's
really
interesting
because
I
I
went
through
similar
exercises
just
an
application
in
financial
markets,
and
we
were
hit
by
many
of
these
questions.
Probably
one
thing
that,
when
playing
with
the
block
size
was
also
the
the
timeout
before
baking
a
block,
so
that
also
was
a
figure
that
we
we
kind
of
change
it
to
to
speed
up
generating
blocks
or
slow
down.
So
that's
also.
B
B
But
the
fact
is
that
if
you
want
to
have
this
system
working
real
time,
you
might
have
to
make
the
tough
decision
and
put
it
put
a
small
enough
timeout
value.
So
you're
assured
that
the
results
of
whatever
smart
contract
that
you're
running
will
be
kind
of
out
after
a
particular
time,
and
if
it's
not,
you
can
define
a
certain
template.
I
have.
I
have
thought
of
that
for
for
kind
of
higher
transaction
through
put
use
cases
for
now,
I'm
I'm
I'm
okay
there.
B
C
B
C
C
E
C
Yeah
one
one
last
comment,
probably
from
my
side
and
I'll
leave
it
to
the
other.
Maybe
it's
is,
I
see
also
a
similar
use
case
for
this,
not
just
for
5g,
because
I
think
one
of
the
questions
was
regarding
where
you
can
apply.
It
is
in
tv
white
space,
which
is
also
another
domain
where
there
is
like
shared
spectrum,
and
there
are
like
users
who
would
like
to
bid
and
utilize
that
spectrum,
which
is
it's
not
as
5g,
but
it's
still
one
spectrum
where
it's
providing
internet
of
internet
access
to
rural
communities
etc.
B
B
So
we
did
play
around
with
ethereum,
and
I
mean
as
as
simple
as
kind
of
deploying
a
smart
contract
with
solidity
and
kind
of
their
remix
that
they
have
having
in
it
at
least
back
then
ethereum
didn't
have
the
the
entire
prize
solution
that
they
have
at
the
moment
when
we
started
with
this
work.
So
really
the
idea
of
having
to
pay
a
gas
price
for
it's
like
having
operators
paying
gas
prices
to
some
other
entities
to
run
their
internal
transactions
with
other
operators.
B
B
I
know
that
you
have
that
the
different
networks
and
you
can
set
up
your
own
networks
and
so
on,
but
it's
still
like
the
idea
of
the
cryptocurrency
being
connected
to
your
to
your
blockchain
was
not
really
appealing
to
us.
Then
I
mean
other
options.
Were
there
by
other
options,
I
mean,
like
the
option
that
I
actually
looked
into.
Was
there
was
a
quarter
art
recorder.
B
There
was
also
another
option,
but
to
be
honest,
I
just
just
essentially
chose
fabric
because
of
the
community
and
the
the
features
that
was
was
around
so
yeah.
I
mean,
I
think,
the
only
other
kind
of
blockchain
platform
that
would
be
worthy
of
looking
into
at
the
moment
would
be
corda
and
also
the
the
inter
enterprise
ethereum
that
I
haven't
had
really
the
time
to
go
and
explore.
C
B
B
I
am
actually
using
a
cloud
network
where
there
is
latency
between
these
vms,
so
these
vms
are
not
really
on
the
same
host,
essentially
necessarily-
and
I
I
think
they're
also,
some
of
them
are
on
different
regions.
Even
so
the
latency
is
there,
so
the
network
latency.
Is
there
at
least
to
a
certain
point.
Of
course,
it's
it's.
The
google
network
that
I'm
using
that's
the
cloud
network.
They
have
fiber
connected
between
all
the
all
their
data
centers.
So
that
is
that
it
that
has
to
be
taken
under
account.
B
You
can
essentially
go
and
check
online,
the
the
latencies
between
different
regions
in
in
in
cloud,
for
example,
right.
So
that's
one
thing.
So
the
latency
between
these
vms
doesn't
help,
especially
that
I'm
using
raft-
and
you
have
the
the
the
the
elections,
the
the
priority
elections
happening
and
so
on,
and
all
these
orders
having
to
communicate
with
each
other.
The
other
thing
that
I've
been
doing
is
that
I've
been
distributing
these
orders
among
different
operators.
B
So
I
mean,
and
then
again
so
the
endorsement
policy
that
we
have
is
there's
the
most
extreme
one.
Is
it's
n
out
of
n,
so
all
the
all
the
peers
have
to
execute
the
smart
contract
so
yeah
I
mean.
Besides
this
I
mean
this
is
an
ongoing
research,
so
we
are
still
trying
to
figure
out
where
the
bottleneck
is.
But
this
is
really
that
the
most
updated
answer
I
can
give
you
for
the
moment.
F
We
will
talk
about
this,
like.
I
think
I
will
ask
you
like
maybe
more
questions,
or
I
I
mean
we
can
discuss
it
like
that,
how
we
can
use
these
kind
of
slicing
and
where
we
can
use
these
kind
of
blockchain
things
over
there.
So
thank
you.
Thank
you
for
your
time,
nema
and
I
think
we
are
over
time
and
yeah
yeah.
A
F
B
Yeah,
just
just
for
the
other
attendees,
of
course,
you're
welcome
to
stay
stay
on
the
call,
but
we
are
now
kind
of
doing
some
internal
things
regarding
the
telecom
sick,
so
yeah
depending
so
I
got
some
input
from
some
people.
They
were
essentially
minor,
kind
of
typos
and
so
on.
I
don't
I
don't
see
as
any
anything
major
so,
and
this
was
really
the
deadline
for
for
the
for
the
latest
recommendations.