►
From YouTube: Hyperledger Sawtooth - Hyperledger Hong Kong meetup
Description
Slides: https://bit.ly/2PR1Oub
Feel free to connect with me on linkedin and happy to see any question I can help too.
https://www.linkedin.com/in/edmundto
A
Hello,
my
name
is
Edmond
and
I
am
volunteer
for
the
hydrologic
community
in
Hong
Kong.
So,
first
of
all,
thank
you
very
much
for
spending
the
evening
with
us.
So
today
we
are
very
pleased
to
have
Anakin
coming
all
the
way
from
the
UK
to
share
with
us
on
our
venture
for
Jake
O'hara.
So
so
before
we
get
started
so
I
just
want
to
share
a
couple
of
housekeeping
notes.
So
first
thank
you
to
our
venue
sponsor
aw.
Yes,
so
Michael
here.
A
If
any
of
you
has
any
questions
on
how
AWS
work
with
other
companies
on
what
they
are
offering
in
terms
of
the
projects
face
is
aspect
to
micro.
Second,
yes,
as
a
community
we
are,
we
are
always
looking
for
volunteers.
So
if
you
are
interested
in
having
the
community,
I
would
love
to
hear
from
you.
So
what
what?
What
we
usually
do
is
to
be
here
a
little
bit
early.
You
know
like
ordering
food
making
sure
our
people
are
checking
and
have
to
record.
You
know
as
West.
A
You
have
to
get
up
pretty
much
testing
and
as
Ivy
to
give.
We
do
have
credible
education
yeah.
So
if
you
guys
are
interested,
please
do
let
me
know
yes,
so
next,
so
the
slides
and
the
recording
will
be
available
on
meetup.com
right
after
this
event,
so
you
probably
don't
need
to
take
some
pictures
and
yeah.
So
next
I
want
to
pass
it
to
Julian
Gordon,
who
is
the
head
of
hyper
ledger
asia-pacific
to
share
fast
on
more
on
the
hydrologic
access.
A
So
we
have
does
that
work.
This
absolutely
wonders
with
the
hackathons
right.
Well,
every
year
we
have
something
called
a
hack
Fest,
which
is
a
similar
name
but
different
where
we
get
all
the
core
developers
come
together.
So
we
have
developers
in
all
these
different
projects,
developing
technology
or
sitting
behind
their
machines
in
different
parts
of
the
world
and
Brian
always
says
it's
great
to
get
them
all
together
at
one
place,
we
do
that
once
a
quarter,
a
pack
do
that
once
a
year
and
next
quarter
is
going
to
be
in
Hong
Kong.
A
B
A
A
A
Next
I
want
to
introduce
our
speaker,
Duncan
Johnson,
what
the
CEO
and
co-founder
of
approaching
technology
partners,
which
is
a
corporate
member
of
hyper
nature,
and
he
is
also
a
member
of
the
hybrid
speaker
Perot,
and
he
also
organized
the
hyper
ledger
scholar
middle,
so
I
would
like
to
next
pass
it
to
Duncan.
So
thank
you.
B
So
my
web
background
I've
been
involved
in
automation
and
operations
for
many
years
elapsed.
Ct,
as
you
know,
take
take
pity
on
me.
Please,
but
one
of
the
things
I
used
to
do
in
the
battle
days
was
I
would
develop
software
and
then
throw
it
over
the
wall
into
the
ops
that
was
DevOps
back
in
the
90s
so
and
they
had
to
either
catch
it
or
drop
it,
but
mostly
try
to
figure
out
what
knows
it
was
with
we've
created,
so
we've
moved
on
and
police
say
a
lot
of
since
then.
B
In
fact,
one
of
my
co-founders
and
co-conspirators
kevin
is
working
with
me
for
the
third
time
and
he
was
the
guy
who
said
to
try
and
catch
this
stuff.
So
when
I
was
CTO
at
his
net,
fixed
income
doing
80s
alternative
trading
systems
back
in
the
day
it
was
his
job
to
try
make
sense
of
whatever
it
was.
B
We
was
a
suitable
application
for
production
so
and
you
learned
the
hard
way
if
you
go
hands
up,
who
has
actually
either
been
involved
in
operations
in
a
production
environment
where
it's
not
necessary
life
or
death,
but
it's
financial
services.
It's
certainly
very
expensive
if
things
go
wrong.
Okay,
if
you
keep
it
with
great
hair,
but
that's
been
prematurely
great
so
anyway,
I'm
gonna
do
a
very
short
introduction
to
us
as
a
company.
What
we're
about
blockchain
primates.
Of
course,
none
of
you
know
what
blockchain
is
I'll
quickly
step
through
that
rest
assured.
B
Having
seen
the
attempt
to
do
this
demo
or
in
fact,
I
think
you
skipped
it
entirely
was
very
wise
because
there's
there's
a
really
nice
sort
of
demo
called
tic-tac-toe
or
knots
and
crosses
and
I
was
struggling
to
get
that
to
work.
So
that
was
not
a
good
experience,
but
I
have
never
worked
in,
please
say
so.
B
Having
still
immense
confidence
in
all
of
you.
With
those
opening
remarks,
the
this
is
Kevin.
Looking
silly
serious
he's,
really
what
I'll
be
doing
today.
A
lot
of
the
work
is
down
to
him
because
when
we
started
out
on
the
stew
with
blockchain
technology
partners,
my
initial
thought
was
to
use
something
called
Apache
Brooklyn,
which
I
know
and
love
having
helped
create,
and
it's
a
top-level
project
within
the
Apache
Software
Foundation
is
really
designed
to
if
you're
familiar
with
cloud
formations.
It's
a
kind
of
generalization
that
idea
that
you
can
create
blueprints.
B
Those
are
essentially
templates,
come
in
and
and
then
use
those
to
deploy,
manage
applications.
But
it's
strong.
Its
strong
suit
is
really
dealing
with
in
many
ways.
Legacy
applications
by
that
I
mean
anything
that
three
days
containerized
platforms,
orchestration
of
said,
said:
containers
platforms
and
the
notion
of
searches
or
micro
services,
so
so
on
the
Kevins
direction
as
chief
strategy
officer,
we've
actually
pivoted
away
from
brooklyn
and
are
now
focusing
on
kubernetes.
B
The
dock
of
the
company
was
given
up
on
top
would
compose
everything
doesn't
believe
that
it's
true,
you
know
if
they
still
offer
it,
but
it's
gone,
and
so
and
what's
important
in
our
view,
is
that
it's
also
it's
also
an
open
source
project
that
was
originally
created
by
Google.
But
it's
now
a
part
of
the
foundation.
So
it's
under
the
cloud
native
computing
foundation,
umbrella,
they're,
lots
of
umbrellas
in
in
the
layers
foundation;
well,
yeah
foundations,
greenhouses
umbrellas
over
loading
zone
project.
B
B
Our
goal
simply
put
is
to
is
to
really
radically
simplify
the
adoption
and
the
use
and
the
value
of
blockchain
technologies,
and
particularly
focus
on
Enterprise
and
hence
the
joining
the
hug
ledger
community.
In
fact,
digression
I'm,
a
repeat
offender
Klaus
off
my
previous
company
joined
in
the
first
wave
of
none
founder
or
core
members
back
in
2016
and
when
I
started
BGP
we
again
signed
signed
the
pledge.
B
But
the
idea
is,
you
know:
bring
bringing
blockchain
out
of
the
out
of
the
science
projects
out
of
the
sort
of
labs
and
making
it
something
is
easy
to
consume
and,
more
importantly,
I
hope,
most
of
your
developers
easy
to
work
with.
So
you
can
actually
develop
applications
because
it's
a
piece
of
middleware
at
the
end
of
day.
You
know,
middleware
is
just
that.
B
Speaking
of
various
events
over
the
next
month,
their
CEO
is,
is
actually
a
lawyer
by
background,
and
so
so
they
decided,
let's
create
a
useful
blockchain
based
application
or
service
name
their
marketplace
for
for
legal
greens.
So
in
the
UK,
if
you
don't
run
a
will,
the
government
gets
everything
English
yeah.
This
should
be
some
sort
of
incentive
to
actually
write
a
will
that
people
are
very
lazy
and
also
days.
You
would
go
to
a
newsagent
and
buy
a
will
from
W
Smiths.
So
the
idea
is:
go
to
a
marketplace,
get
a
document.
B
That's
easy
to
understand!
That's
been
that's
been
curated
fill
it
in
forget
about
it.
It
will
be
looked
after
and
managed
throughout
its
lifecycle.
Well,
it's
only
I
mean
ever,
but
once
so
it's
a
one-time
fire-and-forget,
but
there
are
many
other
contracts
which
which
have
a
lifespan,
be
they
insurance
they'll,
be
they
anything
that
has.
It
has,
has
a
workflow
associated
with
it.
So
look
out
for
Mon
apps
and
their
agreements
network
is
a
great
example,
I
think
of
yeah,
starting
with
a
very
specific
and
very
useful
valuable
service.
B
So
what
do
we
do?
Well,
it's
all
about
the
permission
and
that's
why
we're
here?
It's
all
about
the
permission.
Blockchain
platforms,
we're
creating
one
built
out
of,
as
I've
already
said,
I
purchased,
autism
kubernetes,
certainly
early
on
about
I,
think
you
mentioned
Hortonworks
and
Hadoop,
and
if
you
remember
the
sort
of
whole
history
of
how
it
came
about,
it
was
really
hard
to
wrestle
to
the
ground.
B
It
was
very
complex
and
then
a
couple
of
companies
came
along
cloud
era
was
one
and
Hortonworks
was
another
and
they
actually
created
things
well
within
the
open-source
community
things
I've
got
minbari
that
came
along.
That
gave
you
blueprints
for
a
dupe
and
so
basically
made
it
manageable
made.
It
tractable
made
it
useful
and
then
added
other
services
on
top,
so
this
is
basically
the
same
idea.
So
there's
nothing
new
in
this.
It's
just
the
the
domain
of
discourse
is
is
is
really
blockchain
and
within
blockchain
the
moment
our
focuses
on
IP
address
or
T's.
B
So
what
people
do
to
really
use
this
platform
and,
as
the
server
bullet
point
says,
we're
also
very
happy
to
collaborate
with
them
to
develop
the
operational
model
going
forward.
So
we'll
be
talking
a
lot
today
in
demoing
running
in
the
cloud
in
particular.
Aws,
that's
I
think
a
very
good
first
base,
but
not
everyone
wants
to
stay
in
that
world.
Some
people
want
to
do
a
hybrid
thing,
so
you
know
it's
is
really
making
sure
that
you
understand
and
work
with
and
develop
the
right
operational
model.
B
So
that's
that's
all
about
us,
co-creation
Boeing!
This
is
it.
This
is
something
we
firmly
believe
in
it's
working
together
and
it
sort
of
makes
sense
because
I
come
from
and
have
been
involved
in
open
source
for
over
a
decade
or
more
now
the
notion
of
community
and
collaboration.
You
know
the
source
is:
is
a
given
if
anybody
and
people
still
do.
This
unfortunately
looks
an
open
source
project
and
says
it's
nearly,
but
not
quite
what
I
want
so
I'll
go
and
write
what
I
want.
That
is
that's
kind
of
yeah.
B
That's
that's,
really
not
polite,
because
what
you
should
do
is
say
well
we're
we're
are
how
is
it
different
and
can
I
talk
to
the
people
who
are
working
on
it
and
can
we
have
a
discussion
about
it
or
maybe
I'll
contribute
pull
request
that
says
how
about
we
look
at
it
this
way?
Not
that
way,
so
co-creation
is
doing
the
same
but
more
on
a
business
level.
So
it's
saying
let's,
let's
co-create
a
solution
that,
as
it
says
here,
delivers
a
mutually
valued
outcome.
B
So
a
bit
of
MBA
speak
I,
know,
but
there's
a
serious
point
here.
So
the
primer,
the
slides,
are
here
for
anybody
that
needs
them.
I
pulled
out.
Some
quotes
from
Gartner
I
won't
dwell
on
them,
but
it's
interesting.
Looking
at
the
way,
analyst
view
us
as
a
community
and
blockchain
they're
not
negative,
but
they
are
I
would
say,
has
a
mistake,
which
is
always
the
way
when
you're
ready
so
we're
both
early
and
and
late
in
a
way
because
blockchain
does
concept
has
been
around
forever
but
in
terms
of
its
adoption.
B
Tickly
by
enterprise
was
still
at
a
very
early
stage.
So
so,
but
that
means
it's
the
right
time
to
get
involved.
So
now
is
the
right
time
to
start
a
company
or
get
involved
with
with
with
those
technologies
are
still
nascent
or
not.
You
can
make
an
impact,
make
a
difference,
but
it
will
be
a
while
before
it
becomes
mainstream
technology
and
so
gotta
think
that
that's
going
to
take
sort
of
five
five
or
more
years
and
there
will
be
some
spectacular
disasters
along
the
way
that
there
have
been
a
few
already.
B
There
will
be
more,
you
know,
but
there
will
be
some
successes
and
so,
according
to
them
and
I,
looked
at
the
data
and
it's
that
they're
saying
really
get
sort
of
mid-20s
that's
a
long
time
ago
for
me,
but
the
mid
20s,
that's
in
20,
25
26
and
that's
when
it
will
already
hit
its
stride
and
really
grow
from
strength
and
strength.
So
we've
got
this
real
opportunity
as
a
community
to
get
stuck
in
now.
B
B
God
bless
them
have
done
such
a
bridge,
job
marketing
fabric,
that
people
think
hyper,
ledger
and
fabric
are
one
the
same
thing
they
are
not,
and
the
only
thing
I'll
say
about
the
fabric
is:
if
you're
a
consultant,
you
want
a
job
for
life,
go
with
fabric.
If
you
want
to
work
with
a
modular
sense
of
being
architected
platform,
go
with
sorties.
It's
your
choice,
you
know,
you'll
be
maintaining
this
to
rescue
life,
so
be
careful.
What
you
ask
for
this
is
not
be
record.
B
Serious
point
I
have
raised
this
as
Julian,
because
I
think
that's
the
community.
We
do
actually
have
to
start
addressing
the
question.
What
what
why
are
there?
So
many
frameworks,
five
on
counting?
What
what
do
they
address
or
not
address,
because
we
don't
people
will
make
up
their
minds
our
minds
for
them,
if
nothing
says
so,
I
think
we
owe
it
to
ourselves
and
to
the
community
and
I
would
recommend
that
we
do
the
members
summit
in
in
a
few
weeks
time
to
come
up
with
at
least
a
stake
in
the
ground
and
stuff.
B
Why
we've
got
so
many
of
these
and
what
the
point
is
and
to
be
fair,
I've
made
comparisons
and
also
make
comparisons
when
it
comes
to
seems
like
benchmarking,
but
do
it
in
a
way
that
is
is
is
scientific
and
is
reproducible
because
a
lot
of
benchmarks
playing
around
at
the
moment,
which
are
frankly
rubbish.
So
there
was
a
Deloitte
report.
That's
been
pulled,
thankfully,
because
I
complain
and
said.
This
is
absolute
rubbish
or
words
that
effect,
but
the
reason
it
was
rubbish
was
it
was
comparing
apples
and
pears.
B
It
was
running
sorties
in
depth
mode
and
if
you
read
the
small
print
or
even
lashes
for
quite
large,
it's
normal
sized
print
around
dev
mode,
it
says,
do
not
use
this,
for
any
benchmarking.
Do
not
use
this
in
the
distributed
multi-region
a
benchmarking
exercise
such
as
the
way
that
has
gone,
but
it
caused.
It
cost
me
personally
real
grief
with
one
particular
opportunity
because
that's
what
they
had
so
you
know
I
am
NOT,
going
to
stand
here
and
do
a
start-up
and
not
fight
back.
B
If
we
have
to
deal
with
that
sort
of
stuff,
because
it's
rot,
it's
rot
and
it's
rubbish
and
it
needs
to
be
addressed.
So
I
will
be
raising
this
at
the
members
of
it,
which
is
where
we
can
be
quite
open
and
honest
with
one
another
right.
That's
the
idea.
Yeah
and
people
might
cry
you
know,
but
so
anyway,
let's
get
serious.
Let's
talk
about
the
Sawtooth
and
let's,
let's
go
to
the
man
down
the
man.
B
So
so
what
I'm
now
going
to
do
is
walk
through
and
and
please,
if
you
have
any
questions
jump
in
these
slides
will
be
available
along
the
recording
right
afterwards.
But
down
is
the
sort
of
he's
a
member,
the
TSC.
If
you've
not
met
him,
you
have
to
at
some
point
he's
a
real
live
wire
really
passionate
about
this
knows
his
stuff
is
still
very
much
a
hands-on
sort
of
architect.
So
he's
he's
one
of
the
lead
maintain
as
to
the
core
and
tremendous
guy.
B
So
so
these
slides
are
all
credit
to
him
and
any
any
mistakes.
Any
sort
of
confusion
is
all
down
to
need
on
there,
okay,
so
the
links
there
to
the
code
and
and
so
forth,
and
because
these
slides
were
prepared
before
the
big
announcement,
which
was
a
while
ago,
that
was
like
very
much
timeframe.
So
time
flies,
but
this
was
a
coming-out
parade
for
sorties
I'm,
very
important
that
it
reached
1.0.
B
So
what
was
the
philosophy?
And
this?
This
is
kind
of
important
because
having
been
involved
in
dealing
with
the
large-scale
distributed
systems,
some
of
which
were
not
designed
to
be
large-scale
or
distributed,
and
some
of
which
were
fortunately,
the
successful
ones.
That
is
some
of
the
design
principles
here
were,
let's
start
out
knowing
this
is
going
to
scale
and
have
to
scale
right.
So,
let's
not
think
it's
time
to
do
the
scale
future
I'll
just
axe
here,
yeah,
so
management,
new,
future
zero
scale.
Okay,
that
isn't
it'll
work.
B
Don't
want
to
be
tempted
to
have
a
blanket
ban
on
anybody
running
any
and
not
just
sorta
fabric,
anything
purely
on
their
laptop
unless
they
can
run
a
decent-sized
kubernetes
cluster
on,
in
which
case
they're,
okay.
But
if
what
they're
trying
to
do
is
spin
up
a
few
docker
engines
in
you
know
sorry
docker
images
in
a
docker
engine
that
is,
that
is
frankly
gonna
gonna
confuse
you
it's
going
to
give
you
misleading
results
about
eventual
execution.
B
You
know
this
is
the
antithesis
of
of
high
frequency
trading
platforms,
where
you've
got
to
react
in
9
seconds,
because
you
have
to
understand
that
things
will
take
sometimes
minutes.
There's
always
two
things
were
trying
to
do.
The
same
thing.
I'm
almost
succeeded.
The
other
will
fail.
In
that
sense,
you
have
to
you
have
to
understand
notice
the
caps
theory
which
is
dealing
with
consistency,
availability
and
performance,
but
it's
doing
that
and
it's
blocked
change.
B
So
there's
there's
even
more
current
nuances
around
that
so
having
a
blockchain,
that's
tunable,
so
an
application
or
set
of
applications
is
also
very
important
and
that's
one
of
the
things
that
the
sources
provide.
So
you
can
set
things
like
the
number
of
batches
in
the
transaction,
the
mean
time
between
transactions,
etc,
etc.
There's
only
on
the
size
of
a
block,
but
clearly
you
want
to
be
sensible
about
how
you
treat
these
things.
B
B
I
still
know
every
every
nerve,
every
validator
is
going
to
have
a
complete
copy,
but
what
I
can
do
within
that
tree
is
actually
partition
off
the
TP
for
settings,
the
transition
process
for
settings
from
one
for
a
tic-tac-toe,
from
whatever
identity,
etc.
What
that
really
means
is
in
terms
of
the
way
things
are
operating
on
the
network
level
is
your
basically
ensuring
that
payload
can
be
routed
based
on
just
three
bytes
okay,
so
the
other
thing
I'll
come
on
to
it
in
more
detail,
is
this?
B
Is
the
clear
separation
between
the
network,
which
is
essentially
done?
It
just
does
what
it
does
extremely
well,
but
it
has
no
idea
what
these
transactions
mean
or
the
transaction
families
are
I.
Think
with
transaction
families
just
an
app
through
that
process.
A
business
process
I
should
say,
but
that
allows
for
not
quite
parallel
in
the
sense
of
truly
parallel
but
of
optimized
scheduling.
So
it's
possible,
therefore,
to
determine
at
the
level
of
the
validator
which
is
dealing
purely
with
network
and
is
running
on
0
and
Q
as
a
substrate.
B
Another
good
choice,
because
it's
it's
a
message.
Bus
that
everyone
knows
understands
is
that
you
can
make
intelligent
decisions
and
you
can
be
provided
with
hints
by
the
actual
transaction
processes
themselves
as
to
what
what
is
related
to
what
and
that
gives
you
the
ability
to
optimize
without
actually
destroying
the
integrity
of
the
system.
So
you've
got
that
kind
of
partitioning.
So
you
can
think
of
that
as
partitioning
based
on
domain
of
discourse,
then
you've
also
got
the
ability
to
partition
based
on
where
that
execution
is
taking
place.
B
So
it's
not
the
case
that
everything
has
to
have
execute
on
every
or
every
validator
node
in
the
network
has
to
have
a
copy
of
every
transaction
process,
so
in
other
words,
process
messages
for
every
family.
But
in
fact
it's
something
you
probably
don't
want
to
do.
The
more
you
get
into
organizations
and
multiple
organizations
and
things
like
Kasasa
and
so
on.
There
may
be
stuff
that
you
only
want
to
run
in
a
subset.
So
it's
now
we're
doing
Ben
diagrams
as
well
as
partitioning
those
two
used
wise.
B
They
are
incredibly
powerful,
simple
I
would
say
core
concrete
concepts,
but
used
wisely.
Give
you
a
huge
amount
of
flexibility
and
freedom
and
appeals
to
me
as
the
next
mathematician,
because
it's
a
nice
way
of
sort
of
blocking
and
tackling
and
thinking
of
how
to
divide
up
the
world.
Sorry,
that's
that
point.
Unchain
governess
is
interesting,
because
what
that
means
is
that
you'll
see
there's
a
settings.
Transaction
processor
everything's
done
is
recorded
on
the
chain,
so
every
every
configuration
change,
adding
another
validate
or
whatever
it's
all
there.
It's
all
waters
of
all.
B
At
the
end
of
the
day,
elapsed
time
anybody
know
what
that
is
it.
So
imagine
you
all
asleep
by
now.
Okay,
you
all
sleep
by
now
and
then
at
some
random
point.
One
of
you
will
wake
up,
but
right
now
your
job.
It's
a
handle,
the
next
block
and
the
idea
of
a
proof
of
the
elapsed
time
is
that,
essentially,
that
is
a
fair
mechanism
for
determining
where
processing
takes
place.
Now
it
only
works
information
environment
and
it
can
work
independent
of
X
DX,
which
is
Intel
security
Enclave.
B
So
it's
not
dependent
on
a
hard
way.
Although
you
there's,
a
version
can
use
that
hardware,
but
the
idea
is
that
there
are
within
the
Sawtooth
core
itself
is
checking
to
make
sure
that
there
is
no
bias
so
that
there
is
a
there
is
a
sensible
distribution
of
who
wakes
up.
So
it's
not
like
microwave
that's
sneaky
changed
things,
so
he
gets
woken
up
and
does
more
work
and
it's
crazy
stuff,
not
them,
and
then
they're,
not
picking
on
you
am.
I
sorry
and
thank
you
by
the
way
for
hosting
this
much
appreciate.
B
Success
right
is
the
sort
is
aetherium
integration
set
for
short,
really
powerful,
still
a
work
in
progress,
but
the
idea
here
is
you
can
take
if
you're,
if
you've
been
used
to
working
in
the
steering
world
using
truffle
and
remix
and
and
solidity
and
Zeppelin,
and
things
like
that,
all
of
which
I'm
going
to
say
with
a
straight
face
like
truffle,
especially
a
you
know.
That's
just
kills
me
every
time.
I
say
that,
but
but
now
these
are
tools
that
are
now
the
facto
standard
tools
for
developing
small
contracts
in
that
world.
B
So
the
set
integration
gives
you
the
ability
to
do
that
on
the
sort
is
network.
There's
some
experimentation
going
on
in
the
fabric,
Ward's
try
and
provide
the
same
same
basic
idea,
but
what
you
get
in
sorties
is
you
actually
get
the
the
borough
EVM
integrated
as
a
transaction
process
and
I'll
show
you
a
picture
of
that
in
a
few
minutes
if
I
ever
get
off
this
introductory
slide
so
basic
concept,
because
all
you
really
need
to
know
there
are
clients
they're
important,
is
they
actually
interact
with
things?
B
There
are
transaction
processes
they
do
something
in
response.
So,
and
depending
on
again,
the
domain
of
discourse
messages
will
be
routed
to
them
by
via
the
validator,
which
also
provides
ya
what
they
call
mediated
access
to
the
state,
so
that
states
really
the
global
state
and
in
essence,
the
story
of
how
we
got
to
that
stage.
So
are
you
familiar
with
this
concept?
B
Took
me
widening
my
head
on
it
when
people
had
multiple
trees
and
rose
eyes,
glazed
over
all
the
local
trees
and
in
is
an
efficient
way
of
storing
data,
so
it
can't
be
corrupted.
It's
also
a
very
efficient
way
of
finding
a
way
to
a
particular
part
of
that
tree
and
that
allows
you
then
query
gets
take
back.
The
state
is
being
manipulated
independently
of
the
blockchain,
let's
be
clear,
but
the
blockchain
or
the
block
they've
got
a
block
store
buttons
to
the
chain
record.
B
Every
change
from
the
beginning
of
time
on
that
tree,
so
you've
got
to
have
that
clear
in
your
head.
It
took
me
a
while
to
figure
this
out
so
yeah
as
looking
that's
going.
That's
a
treat
on
a
chain,
no,
no
they're,
two
completely
separate
things.
So
if
you
take
away
nothing
else
for
them
tonight,
it's
just
global
State.
One
thing
the
story
so
far
is
is
really
the
blockchain
itself,
which
will
describe
every
change
made
and
how
that
change
came
about.
So
that's
really
born.
That's
yeah,
that's
kind
of
I!
B
Think
one
of
the
our
hormones
can
be
certain
when
I
was
trying
to
get
my
hair
on
this
stuff
and
sort
of
come
conflating.
The
two
just
causes
yeah,
you
get
a
me
great,
very
quickly,
so
fleshing
this
out
as
you'd
expect
from
this
pattern.
All
that
really
happens
that
sort
of
makes
life
easier
is
that
there
is
a
REST
API
that
sits
in
front
and
the
standard
sort
of
comms
is
product
bus.
B
B
There's
the
block
management,
which
includes
the
block,
store
and
then
there's
a
state,
and
this
is
where
I've
not
been
cleared
before
you
can
see
the
separation
between
them
what's
happening
at
the
network
layer,
which
is
all
about
routing
opaque
messages
around
what
the
client
is
asking
for
them.
What
the
kinds
of
compressor
is
doing
response
to
those
requests,
okay,
really
simple,
really
clear
and
what
works
a
treat
so
then?
Finally,
you
know
I
went
on
this.
You
can
study
this
at
your
leisure,
but
but
there
is
this
notion
of
parallel
or
optimized
scheduling,
firth.
B
It
was
called
optimized
or
intelligent
sharing.
That's
going
on
that
allows
you
to
sort
of
avoid
bottlenecks
as
far
as
possible,
and
what
each
of
these
transaction
processes
represents
is
is
essentially
a
distributed
application,
or
rather
instances
of
the
transaction
process
adopted
around
the
network.
Provide
you
with
that
distributed
application.
Okay
and
the
clients
can
come
in
from
anywhere
again.
If
you're
testing,
you
know,
make
sure
clients
come
in
from
anywhere.
B
So
when
I
get
around
to
doing
the
demo,
you'll
see
us
using
a
load
balancer
in
this
case
and
ELB
in
front
of
the
rest
services
for
you
to
the
ballot
is
to
distribute
mobile
with
when
we're
firing,
requests
into
tests
and
again
it's
101
QA,
don't
just
sit
there
and
fire
things
with
one
validator
a
network
of
a
hundred
or
whatever.
Because
that's
that's,
that's
not
the
way
it's
going
to
work.
B
Unless
you
have
one
client,
because
you've
written
really
rubbish
application,
then
there's
one
person
using
which
kisses
that
it
is
a
fair
way
of
modeling
that
but
anyway
same
as
fabric,
there's
a
there's
a
high
degree
of
choice
of
language
that
you
can
use
for
the
both
the
client
and
the
translation
browser.
And
that
again,
is
because
there
was
this
very
clear.
Separation
was
being
the
network
which
you
really
don't
need
to
know
what
it's
implemented
in
the
moments
Python.
B
But
there's
a
lot
of
reworking
going
on
because,
as
you
would
expect,
as
this
thing's
been
words
a
lot
more
people
there's
been
sort
of
a
certain
amount
refactoring
which
I'll
mention
again
in
a
second
but
also
reworking
at
some
of
the
core
components.
But
if
you
are
thinking
of
this
from
a
developer
point
of
view,
there's
no
mandate,
you
know
as
long
as
you
follow
the
guidelines,
for
you
know,
developing
the
transaction
family
specification
and
then
developing
the
transaction
procedure
to
understand
that
set
of
messages.
B
Then
then
you
knock
yourself
out
and
then
finally
Seth
as
I
said
before
there
you
go
it's
it's
a
transaction
process
with
a
difference
in
that
it
also
provides
this,
and
this
is
what
people
expect
in
theory,
there's
a
there's,
a
an
RPC
because
music,
it
will
understand
and
will
happily
talk
to
you
or
you
can
talk
to
it.
I
should
say
through
your
favorite
tool.
So
from
a
CI
CD
point
of
view
you
can
carry
on
ask
before
there
are
a
couple
of
not
so
much
gotchas,
but
there
are
things
that
there's
no
gas.
B
Okay
and
some
of
the
detail
and
I
said
up
the
lapped
CTS
just
set
us
on
principle.
There
there
was.
There
are
some
things
that
have
to
be
provided
where
services
that
don't
quite
map
so
blocked
info
is
one
of
those,
but
the
guys
are
working
on
and
there's
a
good
core
set
of
maintain
is
working
on
making
sure
that
that
doesn't
violate
the
the
compatibility
okay
for
the
end
of
the
day.
B
That
I
think
is
it's
not
a
one-off
science
project
by
somebody
trying
to
hack
it.
It's
actually
a
part
of
the
sort
tooth
family
and
it's
it's
a
good
example,
reusing
stuff
from
another
hyperlocal
project,
in
this
case,
burro
from
bone
out,
so
I'm
trained
government
configuration
stored
if
it
changes
that
the
change
is
stored.
All
of
this
you
can
actually
see
when
you
first
start
up
Salty's
network
there's
a
lot
of
activity
blocks
are
created
and
also
some
other
good
stuff
going.
But
I
didn't
do
anything.
No,
you
didn't
well,
yes,
he
did.
B
B
The
notion
of
pluggable
consensus
protocols
really
really
important
and
do
not
try
this
at
home,
but
it
isn't
very
possible
to
change
the
consensus
protocol
on
the
fly.
I'm,
not
sure,
there's
a
good
use
case,
but
this
leaps
out
at
me.
But
the
point
is
it's
configurable
like
everything
else,
so,
joking
aside,
there
might
be
a
rationale
for
these
updating
a
protocol.
You
know
et
cetera,
but
that's
that's
really
a
heavy-duty
discussion
and
that's
really
down
in
the
weeds
of
how
do
we
operate
this
over
not
hours
minutes
days,
but
months
years,
decades
Sofia.
B
So
there
are
a
number
of
since
future
consensus
options.
They're
actually
already
available,
if
you're
working
with
the
sort
of
the
master
that,
if
you're
ahead
in
the
business
sorties
project,
because
they're
already
available
for
you
to
experiment
with
1.1
is,
is
due
out
imminently,
I'll
say
so
with
me.
So
far,
it's
great.
It
is
and
so
much
easy
to
understand,
but
pretty
much
any
other
framework.
B
B
We
think
it's
easy
enough
to
understand
that
people
should
have
a
good
grasp
of
the
operations
in
terms
of
how
to
deploy
at
least
a
fairly
standard
single
kubernetes
cluster
config,
but
then
really
split
it
into.
You
know
more
detail
on
the
ops
for
people
that
want
that
more
detail
on
the
application
development
and
how
to
approach
that
for
people
that
want
that.
Well,
because
we're
trying
to
split
things
in
the
middle
but
I
think
everyone
needs
an
appreciation
of
the
other.
B
But
if
your
day
job
is
developing
apps-
and
it's
probably
good
idea
to
spend
more
time
on
that-
oh
it
hasn't
so
as
I
sort
of
said
before,
there's
clients
initiate
I
mean
who's
done
of
you
wanted.
Programming
who
understands
meant
to
some
message
is
already
the
same.
Invoking
a
method
is
sending
a
message:
yeah
makes
sense.
This
is
all
this
is
really
so
so
yeah
you're,
essentially
there.
B
If
there
a
set
of
public,
the
defined
method
calls
which
are
ready
messages
in
this
world
or
transactions
to
come
in
inbound
validated
will
route
them
to
probably
by
default,
probably,
but
not
always
to
a
transaction
processor
local.
If
it
has
one
that
is
in
the
right
family,
if
that's
loaded,
it
might
actually
fan-out
work
across
the
network
and
bear
in
mind.
This
is
back
to
what
I
was
saying
earlier.
These
messages
are
coming
in
from
clients
notionally
or
potentially
attached
to
every
validator.
B
So
so
it
makes
sense
where
you
can
to
process
at
least
that
initial
production
locally,
that
doesn't
mean
you're,
suddenly
not
that
issue.
The
application
just
means
you're
optimizing
performance,
but
the
idea
is
that,
as
allied
with
that
is-
and
this
is
the
critical
thing
and
the
hardest
thing
to
get
right-
is
the
data
model
itself?
B
So
what
are
all
these
method
calls
all
these
messages,
all
these
transactions
do
to
affect
or
change
the
state
and
I'll
give
you
the
example
in
Tic
Tac
Toe,
where
obviously
one
of
the
methods
or
one
of
the
messages
or
transactions
store,
the
same
thing
is
create
a
new
game,
and
so
the
transaction
processor
says
alright
happy
to
do
that.
As
long
as
various
things
are
true,
it's
a
game
with
a
new
name,
so
I'm
not
trying
to
create
the
same
game
twice
and
the
person
or
the
identity
of
the
user.
B
The
assertion
is
that
jack
or
Jill
and
the
example
we'll
look
at
exists
and
has
the
credentials
to
actually
create
a
game
and
so
on
and
so
on,
and
then
and
then
it
will
do
things
like
make
sure
that
Jack
doesn't
do
two
moves,
because
we
do
two
moves
playing
notes.
Of
course,
you're
going
to
win
two
in
a
row.
B
So
it's
a
nice
example
because
it's
something
which
I
think
is
has
been
played
for
centuries
under
different
names,
and
so
it
gives
you
a
good
way
of
understanding
how
to
build
a
transaction
process,
the
transactions
in
the
family
and
also
believe
by
the
way
it's
how
we
all
learn.
We
take
something
apart
and
then,
hopefully
put
it
back
together.
I
did
this
badly
when
I
was
a
kid,
we
were
one
car,
family
and
I
thought
I
would
fine
tune
my
dad's
car.
So,
with
a
couple
of
mates
we
took
the
engine
out.
B
Although
I
am
a
firm
believer
that
most
men
was
only
make
sense
once
you
understand
the
thing
itself,
so
you
yeah,
there
are
great
people
that
can
write
documentation,
but
it's
so
hard
to
do
to
get
it
to
the
point
where
somebody
can
actually
learn
through
that
process.
Rather
than
saying.
Oh
now,
I
understand
have
words
because
I
figured
out
another
way
now
I
read
the
manual,
it
will
make
sense
so
anyway,
so
I've
kind
of
said
this,
the
twiddles
I
would
say
saying:
twiddles,
smart
contracts
immediately
puts
you
into
this
sort
of
hole.
B
What's
a
small
contract
debate
and
I
think
you
know
what
is
the
small
Condor
as
far
as
I'm
concerned
is
something
that
runs
on
its
theorem,
okay.
So
as
far
as
I'm
concerned,
that
is
the
definition,
but
these
are
quote
or
twiddles
smart
contracts.
What
they're
realize
is
it's
a
business
logic,
hopefully
fairly
complete
things.
You
forgot
about
the
page
cases
that
you
haven't
sort
of
accounted
for,
but
you
know
the
end
of
day.
B
Engineers
induction
right
engineers
mostly,
can
only
have
Karen
three
things
in
the
head
at
one
time,
so
you
know,
do
you
have
a
hood
mention
his
induction
yeah?
Well,
that's
something!
It's
three!
It's
it's!
That
is
your
inductive
proof.
You
know
it's
true
for
N
equals
zero
one
two.
It
must
be
true
for-
and
the
point
is
the
back
to
the
point
I
made
earlier
about
the
bio-data
and
the
network
being
incredibly
simple,
but
just
good
at
what
it
does
all
you
can
do.
B
As
a
transaction
processor
is
say,
hi
I'm
here
and
provided
you
are
accepted,
allowed
to
be
running
on
that
validator
then
you're
connected
and
then
all
you're
actually
going
to
do
is
then
be
asked
to
process
a
transaction
so
apply.
And
then
all
you
can
do
in
return
is
is
get
upset
somebody
somewhere
in
that
tree,
okay,
you
then
yeah.
B
This
obviously
is
the
key
in
terms
of
have
you
thought
of
everything.
Have
you
have
you,
as
I
said
before
account
for
every
every
edge
case,
but
in
principle
that's
all
you
need
to
know.
Is
you
know
the
interaction
with
the
network
beyond
the
network
is?
Is
extremely
straightforward,
it
should
be,
and
then
it's
the
Devils
in
the
detail
in
terms
of
what
apply
then
actually
does,
but
all
of
that
is
only
in
the
context
of
the
transaction
processor
and
the
state.
It
is
managing
the
tic-tac-toe
game,
it's
state,
etc.
B
B
A
B
We
were
interested
more
Edmund
did
and
in
the
in
the
spirit
of
co-creation
aka
I,
don't
say
that
we
would
like
Edmund
to
try
and
get
the
supply
chain
example
to
work
on
a
instance
of
the
sorted
Network
which
we're
going
to
create,
or
he
will
create,
and
the
reason
that's
important
is
because
it's
a
good
test
of
have
we
done
everything
right
it's
to
give
somebody
else,
the
keys
to
the
king
of
Mencia.
They
can
actually
specify
essentially
add
that
transaction
processor
to
the
mix
and
the
manifest
and
so
on
and
so
forth.
B
So,
looking
forward
to
doing
that,
so
as
down
would
say,
check
it
out,
check
it
out,
get
involved
whether
that's
improve
the
documentation,
definitely
would
be
helpful,
whether
it's
testing
it
whether
it's
getting
to
the
guts
of
it.
It's
all
there.
So
dancers,
thanks
for
listening,
Sherwood
who's
here.
B
So
next
slide.
Let's
talk
about
taking
a
lot
of
this
and
making
it
if
you
like,
tractable
with
reusable
repeatable
by
mere
mortals,
like
myself,
I'm
the
idiot
that
makes
sure
everything
we
do
is
improve.
That's
what
it
means.
It
means
something
in
the
company
who
will
invariably
break
things
through
no
fault
of
their
own,
just
because
they
will
stumble
into
things,
because,
but
not
the
engineer
that
wrote
it.
That
knows
never
do
that,
because
they'll
just
guarantee
to
do
that.
B
B
Is
something
which
is
signed,
sealed
and
delivered
will
work?
We've
built
the
the
docker
images
that
are
pulled
down,
they're
pulled
down
from
a
well-known
bunker,
hunt
organization,
etc,
etc,
etc,
and,
if
you're
a
big
enough
and
bad
enough
enterprising
one
all
that
on-prem.
Yes,
we
can
give
it
to
you,
but
in
reality,
are
you
better
off
at
these
starting
and
looking
at
their
staff
and
working
with
it
in
the
cloud?
B
Because
it
won't
miss
words
and
then
the
other
thing
is
besides,
that
is
a
unified
user
experience
and
what
do
I
mean
by
that
you'll
see
me,
but
what
I'm
going
to
demo
is
what
we're
not
ever
recommending
anybody
ever
ever
I
mean
ever
does
in
production,
because
it's
full
of
risk.
Why?
Because
it's
me
actually
water
but
I'm,
going
to
do
that
to
unpack
same
as
I
was
saying
that
it
unpack
what
is
going
on
behind
the
scenes.
B
So
you
get
a
real
I
hope
sort
of
granular
sort
of
tactile
feel
through,
but
the
idea
is,
as
part
of
that,
you
have
a
unified
experience
both
in
terms
of
specifying
how
you
want
the
network
to
look.
If
you
want
a
rough
approximation
to
what
it
might
look
like
look
at
the
Cluedo
example,
which
is
actually
running
in
the
end
of
this
marketplace.
So
please
is
a
consensus
company,
and
so
they
focus
on
standing
up
its
theorem
for
configuring
not
standing
it
up
there,
configuring
I
think,
but
the
up,
but
what's
nice
about?
B
Is
they
give
you
a
very
simple,
step-by-step
way
of
you
know
rolling
out
an
environment
that
you're
happy
with,
and
so
you
know
we
have
a
single
and
it's
all
about
that
radically
simplifying
the
enterprise
adoption
of
this
technology.
That's
in
a
way
we
there
will
be
the
usual
starter,
professional
and
enterprise,
which
means
the
price
goes
up
as
you
go
along,
but
it
means
also
that
you
get
more
in
the
way
of
an
SLA
around
management
and
operations
and
support.
But,
crucially,
you
can
start
with
a
credit
card
and
an
Amazon
account.
B
And
so
you
can
start
small
pace,
you
go
ask
us
questions
on
slack
or
whatever,
probably
reach
out
here,
I
guess,
as
I've
discovered
when
I
create
the
slack
Channel
Island.
Why
did
you
do
that?
Duncan?
It's
so
easy!
Just
do
we
tell
the
the
marketplace
is
a
I'm?
A
huge
fan
of
this
was
quite
involved
with
this
class
off
with
particular
with
Dave
McCann
who's.
The
VP
for
a
diverse
marketplace
on
the
Service
Catalog,
so
class
off
was
more
involved.
B
I
would
say
on
the
Service
Catalog
side,
which
is
really
take
some
slice
of
the
marketplace
and
curate
it
for
enterprise
and
integrate
with
things
like
stuff
is
now
and
so
on.
So
you
get
the
whole
ITSM
experience,
because
you
know
large
companies
still
want
IT
Service
Management
over
about,
but
but
what
what
they've
achieved
in
a
few
short
years
is
they've
made
the
marketplace
the
concept,
but
most
of
the
reality,
the
the
default
channel
for
distributing
software.
So
you
don't
download
it,
you
run
it.
B
You
spin
it
up,
and
even
all
of
course,
have
kind
of.
You
know
agree
to
this,
which
took
a
while
to
do,
but
the
idea
is
you
go
to
the
marketplace
and
whether
it's
a
simple
Am
I,
you
spin
up
an
instance
and
you're
done
or
whether,
in
our
case,
we
spin
up
an
instance
and
that's
kind
of
if
you
like
the
scene.
That
would
then
allow
you
to
then
deploy
various
networks
or
whether
it's
a
site
offering
it's
all
there,
and
and
now
it's
got
the
point
where
it's
integrated.
B
So
as
a
third
party
like
ourselves,
you
can
also
do
integrated
billing
through
it.
So
that
means
that,
from
a
customer
point
of
view,
its
freshness
and
from
a
provider
service,
it's
also
crucial.
It's
kind
of
a
win-win
I,
don't
it'll,
be
there
so
I'm
gonna
show
you
what
it
is.
I
suppose
this
is
really
hard
to
understand
right.
So
that's
how
difficult
blocked
in
really
really
is
ok,
so
you
know
I've
rather
got
to
explain
this
in
trennis
detail.
B
B
Use
exactly
this
pattern
in
in
real
life
in
insurance,
so,
rather
than
disrupt
everything
safe
up,
throw
everything
away,
I've
got
to
start
from
scratch,
which
is
nonsensical.
They
said
no,
let's
use
blockchain
for
what
it's
been
designed
for,
which
is
to
record
critical
data
in
an
immutable
fashion.
B
True,
just
it's
by
accident
or
design
yeah
errors
creep
in
so
this.
This
is
a
good
example,
and
then
there
are
then
there
are
ways
of
downstream
taking
information
out
of
the
chain
so
rather
than
clearing
the
chain,
which
is
actually
quite
hard
to
do
in
hardened
sense
of
tedious,
there's
nothing
wrong
with
actually
doing
what
insulted
terms.
It's
called
a
ledger
sack
with
let
your
sink.
That's
why
in
C
does
exactly
that.
B
A
B
To
make
sure
that
data
centers
really
are
performing
to
the
PUA
that
they've
been
told
they
have
to
perform
1.3,
but
I'm
not
fudging
that
they
do
not
there.
Any
data
center
hosting
provider
would
ever
do
that,
but
in
the
rare
just
to
eliminate
that
potential,
you
can
do
it
on
the
blockchain
sip,
for
example.
So
you
go
from
simple
to
much
more
complex,
so
all
right
demo,
so.
B
What
can
I
say?
The
good
news
is
we're
using
standard
tools,
that's
very
good
news,
so
we're
not
saying
welcome
to
our
world.
Now.
Here's
a
whole
bunch
of
new
stuff,
you've
gotta
get
head
around
Colin
Cassady
said
earlier.
You
won't
actually
probably
most.
You
needs
no
or
cops
and
Cooper
control
Cooper
CTO
having
pronounced
it
really
is
about,
but
they
are
standard
their
standard
tools
and
that's
important
so
in
the
case
of
cops,
has
up
his
third
of
course.
B
A
B
It's
extraordinary
because
when
I
first
looked,
it
I
thought
that's
interesting,
but
what
I
looked
it
again?
I
thought
wow,
that's
really
interesting
and
when
Kevin
said,
hey
we're
using
I'm
going
wow,
that's
extremely
interesting
and
then
I
looked
at
the
rate
of
progress,
the
number
of
contributors
and
we're
on
1.9
point
2,
which
was
released
at
the
end
of
July,
July
24th,
that's
out
of
date,
quote-unquote
there's
one
point:
10.0
came
out,
10
11,
maybe
12
days
ago
and
I'm
sure
it's
there's
another
one.
B
By
now
and
there
are
like
hundreds
of
people
contributors,
so
you
know
people
really.
What
are
they
saying?
What
are
they
telling
us?
Cooper
knows
has
reached
the
point,
critical
tipping
point
where
it
has
to
be
managed,
and
so
that's,
where
dinner,
create
the
utility
for
exactly
that.
So
this
is
about
spinning
up
the
Cuban
editors.
It's
got
nothing
to
do
with
anything
and
then
and
then,
and
then
you
control
or
coop
CTR
is,
is
the
way
you
interact
with
said
cluster.
So
without
further
ado.
B
I
just
sit
quietly
so
I,
don't
know
and
doesn't
want
to
do.
That
was,
like
confess,
I
confess,
because
killing
things,
but
it
now
has
a
enamel
so
that
it
can't
be
deleted
by
the
you
did
those
things
in
deep
proof,
but
the
reason
I'm
using
that
is
that
I
have
an
accountant
that
also
I
think
Kevin
is
give
me
an
account,
so
he
can
actually
keep
track.
What
I'm
doing
and
I
can't
do
too
much
damage
and,
as
you
can
see,
he's
running
a
kubernetes
cluster
himself.
B
B
B
B
B
B
My
key
to
that
serve
we're
controlling
access
to
this
and
then
update,
updates
very
powerful
cops
update
cluster
name
of
cluster.
Yes,
actually
does.
If
you
don't
do
yes,
it
just
does
a
drive
around
and
tells
you
what
it's
going
to
do,
but
in
this
case
there
is
no
cluster
or
there
was
no
cluster.
So
update
meant
not
just
incrementally
changing
to
conform
to
the
definition,
because
you
know
that
the
definition
once
it's
stored
but
actually
said
wow
I'm
miles
from
having
this
cluster
there's
no
master.
B
There's
no
there's
no
notes,
so
I've
been
a
better
bring
up
the
whole
thing,
which
is
what
it's
now
doing.
So
if
I
refresh
here
well
yeah,
it's
spun
up
a
bunch
of
machines
which
are
now
my
machines
for
my
cluster.
So
none
of
this
is
hard,
but
our
idea
is
just
to
make
it
even
easier
and
so
that's
happening
in
the
background.
B
A
B
Again,
we'll
make
us
available
very
shortly,
but
we
have
definitions.
Service
definitions
in
kubernetes
speak
for
a
number
of
core
building
blocks
and
where
it
says,
type
equals
load
balance.
That's
that's
because
we're
using
Adobe
AIR
support
will
equate
to
an
e
lb
plastic
load
balancer.
The
reason
for
that
is
because
it's
going
to
provide
a
low
balance
access
to
the
those
individual
rest
registers,
which
is
useful
for
kind
of
testimony.
B
B
Those
metrics
are
everything
from
low-level
stuff
through
to
the
actual
save
the
world
number
of
transactions
number
of
blocks
created
that
kind
of
thing
back
any
backlog,
that
sort
of
thing
and
then
those
those
then
feed
into
three
stateful
sets
so
we're
going
to
have
instance,
one
of
monitoring
cuz.
We
only
want
one.
B
A
A
B
B
See
it
was
much
softer
spotting
up
right,
so
basically,
what
it's
saying
now
is
not
that
was
easy
bit,
but
it's
now
set
up
a
kubernetes
which
is
good,
so
control
to
see
you
at
that
and
now,
if
I
quickly
flip
back
to
here,
I
won't
put
it
into
slide.
Marry
the
bus.
Others
mistake
I'm
now
going
to
prove
there
is
no
man
behind
the
curtain
with
that
expression.
B
B
A
B
B
A
A
B
B
B
If
you
look
at
the
count,
the
others
got
ten
out
of
ten.
This
we've
got
five
out
of
ten.
What
it's
saying
I
think
is
I'm
going
up
to
the
docker
hub
and
we're
going
to
the
public
docker
hub
I'm,
trying
to
pull
down
a
particular
dollar
image,
and
it's
failed
our
way
through
that.
So
it's
trying
to
stop
a
component
and
it's
failed.
So
it's
done
a
couple
of
restarts.
It's
now
got
to
seven.
This
is
a
bit
lumpy.
B
Thing
about
the
kubernetes
is
it's
sorted
out
if
you
notice
it's
now
right
so
so
these
are
the
kinds
of
things
that
are
designed
to
strike
terror
into
the
heart
of
the
demo
guy,
but
are
actually
quite
good
because
in
a
sense
it's
it's
it's
what's
good
about
the
kubernetes
start.
Is
these
things
are
designed
not
to
fail
permanently
but
designed
to
be
able
to
recover
gracefully
for
failure?
So
if
that's
true
as
well,
then
yes,
so
now
we
can
see
all.
B
So
we
won't
look
at
will
and
explore
the
network
anymore.
I've
done
that
so
tic-tac-toe
as
I
said
before
it's
a
it's,
a
good
I
think
a
very
good
example
of
work
on
work
from
because
you
kind
of
know
what
it
should
do,
there's
nothing
to
explain
there,
but
it
allows
you
to
unpack
how
this
thing
works.
So
the
the
fun
thing
is
you've
got
to
actually
run
another
neat
feature
which
is
run
a
salty
shell.
What's
that
going
to
do
no,
not
a
huge
amount,
but
it's
it's
very
helpful.
A
B
B
A
B
A
A
B
A
B
A
A
B
A
B
A
A
B
A
A
B
A
tune
or
something,
no,
it
will
tell
you
the
state,
the
state,
so
the
stable
would
be
tie
if
it's
tie
or
win
lose
so
yet
so
it
and
he
knows
one
there's
nothing
else
to
do
it
when
it's
in
it
when
that
games
in
the
past
day.
Yes,
so
you
know
we
can
carry
on
syringes
and
Dutch
means
I'm
happy
now
it
must
work.
But
now
maybe
we
can
carry
on
so
Jack
I
guess
it's
gonna
go
seven.
A
B
B
B
A
B
Be
on
a
panel
talking
about
something
the
we'll
be
talking
about
some
of
the
challenges
of
it.
What
does
it
mean
to
go
from
proof
of
concept
to
production?
What
are
the
challenges?
It's
not
just!
That's
not
just
that
sort.
Success,
yeah
previa,
the
whole
bunch
of
people
are
talking
about
their
experiences,
of
whom
own
acts
is
probably
one
of
the
most
experienced,
because
learners
have
been
running
burrow
for
four
years
using
conveyance
the
open
source
of
some
ik
Europe,
which
actually
I
can
walk
to.
For
my
current
home
in
Edinburgh.
B
B
So
I
would
say:
that's
you
know
and
that's
a
chance
for
everybody.
So
the
medicine
is
the
internal
private.
You
know
everyone
talking
under
the
Chatham,
House
or
club
rules
trying
to
sort
of
make
sure
we're
happy
with
the
overall
direction,
but
the
global
form
is
the
is
really
inviting
the
world
to
come
and
find
out
more
about
what
hybrid
community
is
doing.
B
B
B
However,
it
says-
and
this
is
really
really
important-
joking
aside
here
within
three
to
five
years
response
believe
China
will
overtake
the
u.s.
shifting
the
center
30
center
of
inference
or
grounded
will
actively
from
the
US
and
Europe
something
to
think
about,
and
a
lot
of
the
reason
why
I'm
here
and
intend
to
stick
around
so
with
that
thought
hope.
This
was
interesting
and
useful.
Sorry
I
went
on
a
bit.
A
B
B
Then
you
know
you
can
imagine
people
having
control
of
their
own
clusters
quite
very
smoothly,
use
the
same
tools
so
that
you
know
hopefully
there's
some
coordination,
but
you
wouldn't
have
a
cluster
of
N
and
plus
of
em,
where
they're
fully
meshed,
because
standard
networking
and
wide
area
network
them
say
that
you've
got
you
nominate
some
some
pair
or
more.
These
two
represent
sort
of
the
I,
don't
think
gateways,
that's
an
old-fashioned
term,
but
are
essentially
the
the
ways
in
which
you
would
mesh
those
together.
So
there's
a
lot
more
besides.
B
B
We
have
to
recognize
because
networking
has
been
around
for
time-
there's
a
lot
we
can
learn
from
from.
You
know
our
peers
in
the
in
the
operations
and
networking
space
on
how
to
do
this.
So
so,
what's
blockchain
and
things
like
sort
of
in
fabric
might
be
relatively
new.
There
are
some
best
practices
that
we
need
to
follow
in
the
case
papers.
B
You've
got
this
sort
of
notion
of
a
landing
zone,
so
you
know
that's
an
example
not
within
the
context
of
blockchain,
but
it's
a
it's
a
it's
a
well-defined
way
of
a
company
connecting
to
the
cloud
and
so
I
expected
to
see
best
practices,
but
it's
from
you
or
the
other
hyper
cloud
providers
providing
us
with
some
of
that.
So
at
the
moment,
if
I
go
use,
a
template,
that's
provided
by
ADA.
B
It's
it's
it's
bringing
up
a
sort
of
basic
kind
of
development
environment
for
people,
but
that's
not
what
you'd
once
run
in
the
future.
So
so
I
think
you
know
its
ability
to
scale
as
long
as
you're
intelligent
about
how
you
deploy
it's
one
of
the
attractive
features
and
its
simplicity
and
there's
a
small
number
of
moving
parts.
As
you
could
have
seen
those
you
know
pretty
it's
just
the
validators
and
the
networking
running
at
one
level
and
then
whatever
applications,
whatever
translation
process
founders,
you
want
to
run
above
that
so
so.
B
Scaling
a
cluster
and
then
there's
actually
scaling
an
application
of
kubernetes
application.
So
sorting
is
an
application
running
and
designed
the
way
that
we
put
this
together
so
to
play
nice
with
kubernetes.
So
you
can
scale
the
number
of
validators.
The
limiting
factor
is
that
the
number
of
validators
should
not
exceed
the
number
of
nodes
in
the
cluster,
because
you
do
not
want
valid,
it
is
co-located
on
the
same
underlying
infrastructure
and
that's
not
because
you
couldn't
make
it
work
is
because
why
would
you
do
that
in
the
kind
of
cloudy
world?
B
Means
you
always
get
a
validator
every
time
you
essentially
bring
up
a
node
in
a
kubernetes
cluster.
You
automatically
get
the
validator
the
challenge
with
that
I
think
that's
reasonable,
because
you
know
in
a
sense-
and
this
is
nothing
to
bear
mine
we
envisage
rightly
or
wrongly,
that
these
clusters
will
be
generally
speaking,
dedicated
to
managing
the
blockchain
Network.
Why?
Because
again
cloud
yeah,
why
would
I
run
some
other
random
application
on
the
same
cluster?
I
can
run
it
on
another
cluster
because
we
know
from
experience
and
certainly
in
Mike's
racing
in
the
past
yeah.
A
B
B
We
do
not
want
every
validator
to
run
every
known
TP,
so
there's
all
work
to
be
done
to
make
real
that
Venn
diagram
I
talked
about
where
you're
actually
deciding
or
determining
what
TPS
can
run
on
one
validators
and
that's
still
work-in-progress
for
one
of
us
but
yeah
I,
think
I
think
you
can
show
the
two
playing
nice
in
that
yeah.
There's
nothing
wrong
with
getting
out
the
cluster.
B
We've
chosen
to
focus
on
kubernetes
because
it's
as
portable
as
it
gets
and
the
early
work
that
was
done
with
dr.
composed
by
both
fabric
and
sorties,
has
severe
limitations
and
is
only
really
doing
it
kind
of
almost
like
a
laptop
environment.
So
it's
it's
kind
of
horrible,
so
I
think
yeah
I
would
say
we're
all
almost
all
connected
these
days.
Almost
all
the
time
so
I
would
tend
to
say
you
know,
establish
an
environment
as
your
development
environment.
Use
that,
if
you
really
want
to,
you
can
obviously
shut
it
down
but
you're
talking.
B
Hence
you
can
set
it
so
that
actually
you
lose
max
prices,
so
it
goes
to
spot
instances.
That's
something
else
that
Kevin's
done.
We
turned
that
off
the
demo
because
I'd,
like
my
instances
to
come
up
and
not
hang
around
waiting
because
the
spot
price
has
to
be
higher
than
we
were
expecting,
but
so
there
are
lots
of
other
ways,
kind
of
being
judicious
in
your
use.
But
but
you
know,
if
you've
ever
done
something
on
your
laptop
and
then
shut
the
laptop
there
then
open
it.
It's
a
pretty
disastrous
nominee
versus
you
know.
B
Just
getting
kicked
out
of
a
session
and
reconnecting
so
I
think
I
think
you
know,
treat
distributed
systems
seriously,
make
sure
the
environments,
even
the
development,
even
support
the
virus
are
a
reasonable
approximation
to
what's
going
to
be
happening
in
the
real
world
and
the
generous
peak.
That
means
multiple
machines,
at
least
yeah.
A
B
The
moment
we're
focusing
on
just
getting
things,
stay
warm
one,
but
that
will
be
the
next
next
iteration
is.
What
we
want
to
do
is
put
this
out
there
for
people
to
see
and
firm,
with
the
give
us
feedback
on
and
a
couple
people
already
asked
whether
they
can
have
some
access
to
that
before
we
actually
go
completely
public,
we're
fine
with
that
honor
on
a
sort
of
case
by
case
basis,
good.
B
We
want
people
look
it
is
it's
also
prepared
to
put
in
the
time
and
and
say
yes,
so
when
we
get
into
apologies
for
production,
where,
for
example,
you're
doing
consortium
where
it's
logical,
each
member
of
that
consortium
has
their
own
cluster,
then
you've
got
to
start
thinking
more
about
again.
The
networking
consideration
for
doing
that
and
again
in
the
same
way
that
I
talk
about
scaling
off
things,
but
you
had
it's
a
new
feature,
security,
something
you
port
on
after
the
fact.
So
so
you
know
this
environment
that
I'm
using
is
relatively
locked
down.