►
From YouTube: Metadata + DataHub: the Secret to Data Mesh
Description
Stefanie Kechayas, Data Platform Lead from MYOB, shares how DataHub has been critical to her & her team's implementation of data mesh practices during Metadata Day 2022.
Learn more about DataHub: https://datahubproject.io
Join us on Slack: http://slack.datahubproject.io
Follow us on Twitter: https://twitter.com/datahubproject
A
Hello:
everyone,
my
name,
is
stephanie
cachais
and
I'm
the
data
platform
lead
at
myob,
which
is
a
company
in
australia,
a
software
company
and
I'm
very
excited
to
be
presenting
today.
Thanks
for
having
me
I'm
on
the
other
side
of
the
world,
so
I
had
to
pre-record
because
otherwise
it
would
have
been
in
the
middle
of
the
night
that
I
do
this
live.
A
But
it's
lovely
to
be
here
thanks
for
having
me-
and
I
hope
you
enjoy
my
lightning
talk
and
apologies-
I
I
will
have
to
fly
through
things,
because
we
only
have
five
to
ten
minutes
so
I'll.
Try
to
make
it
quick,
okay,
really
quickly
a
bit
about
us
in
case
you've,
never
heard
of
nyob,
so
we're
a
business
management
software
company
we
help
more
australian
and
new
zealand.
Businesses,
start
survive
and
succeed.
A
And
probably
one
of
the
most
remarkable
things
about
us
is
that
we
are
30
years
old
as
a
company
as
a
software
company,
which
is
pretty
remarkable
when
you
really
think
about
the
average
age
of
most
most
software
companies
we've
been
around
for
a
long
time
and
back
in
the
day.
You
know
it
was
primarily
accounting
software,
so
running
running
an
accounting
suite
on
your
on
your
desktop.
A
But
these
days
we
do
much
more
full
end-to-end
business
management
processes
processes
in
our
in
our
software.
We
have
we
organize
ourselves
into
three
verticals
sme,
which
is
smaller
business
full
to
medium
enterprise,
businesses,
so
small.
You
know
solo
sore
traders
and
other
people
like
that
then
enterprise,
which
is
your
bigger
companies,
bigger
businesses,
and
now
we
also
have
a
financial
services
arm
that
provides
a
very,
very
amount
of
financial
service
products.
We
have
about
2
000
staff
across
australia
and
new
zealand.
A
I
work
in,
as
I
said,
the
data
platform
team,
so
we're
a
team
of
around
20
people
who
manage
an
internal
platform
for
data
and
analytics
at
myob,
and
so
we're
responsible
for
all
of
the
infrastructure
and
tooling
and
advice
and
consulting
around
best
practice
data
management
and
giving
basically
all
of
our
users
in
the
business
tools
and
and
ways
to
work
with
data
and
be
data-led
at
myob,
and
I'm
going
to
talk
a
bit
about
the
journey
we've
been
on
with
thinking
about
data
mesh
as
a
concept
and
as
a
philosophy.
A
So,
a
few
years
ago,
some
colleagues
of
mine
and
and
myself,
you
know,
started
learning
about
this.
This
new
kind
of
paradigm,
zamek
dagani,
is
the
author
of
this
idea
and
if
you've
not
read
the
book
or
seen
the
blog
post
at
least
check
out
some
of
the
blog
posts.
It's
some
really
interesting
insights,
and
this
really
spoke
to
us
as
it
spoke
to.
A
I
think
a
lot
of
people
in
the
data
industry
who
are
facing
kind
of
issues
of
scale
and
data
issues
across
across
an
enterprise
and
what
she
was
basically
saying
was
stop
stop
siloing
data
into
these
kind
of
teams
and
think
more
about
data
as
a
product
data
is
something
that
teams
domains,
domain
teams
own
and
architect
themselves
and
then
create
products
around
that,
and
that
in
and
of
itself
is
kind
of
was
quite
quite
good
for
us,
because
we
were
already
going
down
a
domain
driven
design
architecture
at
nyb
in
general.
A
So
it
sort
of
worked
with
that
and
then
this
idea
that
the
infrastructure
as
a
platform
self-service
data
that
that
that
platform,
tooling
in
data
providing
providing
self-service
tools
and
ways
for
others
to
just
start
to
work
with
data
themselves
and
not
be
necessarily
reliant
on
another
team
to
do
their
stuff
for
them,
and
obviously
federation
computational
governance
as
well.
This
idea
that
these
tools
and
processes
that
we
build
into
the
platform
provide
governance
and
data
management
frameworks
alongside
it
and
that's
part
of
you
know
what
we
offer
as
the
platform.
A
There
was
just
a
lot
of
things
about
this
that
that
really
resonated
with
us
and
the
reason
why
it
resonated
was
because
I
think
we
were
going
through
what
a
lot
of
companies
go
through
when
they
start
to
really
grow
and
really
want
to
engage
with
data,
and
that's
where
just
understanding
and
navigating
data
was
getting
really
hard.
We
had
the
data
and
analytics
platform
team
which,
which
is
now
my
team
kind
of
in
the
middle
of
everything,
and
then
on
one
side.
You
had
source
systems
that
really
had
nothing
to
do.
A
That
could
be
done,
even
though
you
know
we
had
good
foundations
of
data
practice
in
that
in
that
new
lake,
but
we
knew
that
it
just
wasn't
getting
used
like
it
should
be,
and
it
just
everything
felt
very
siloed,
and
so
yeah
data
mesh
started
to
really
open
up
a
new
way
of
thinking,
as
did
platform
thinking.
A
So
we
really
started
to
reimagine
how
our
place
in
the
nyb
landscape
and
understand
that
okay,
instead
of
being
a
destination
point
in
the
middle,
this
sort
of
bottleneck
this
this
place
that
people
need
to
go
to
or
this
team
that
people
need
to
ask
for
questions
when
they
have
questions
about
data
instead.
Let's
think
about
ourselves
as
enabling
and
as
just
providing
sort
of
the
underlying
tools
and
platforms
on
which
actually
the
good
con
conversations
need
to
happen
between
data
producers
and
consumers,
and
we
started
to
think
about
the
people
in
the
business.
A
As
you
know,
these
two
main
buckets
you
know
you're
either
producing
data
you're
part
of
the
delivery
teams,
creating
products,
part
of
the
operational
teams
or
you
were
consuming
data.
A
But
some
some
analysts,
engineers
and
data
analysts
actually
are
creating
new
data
sets
themselves
so,
even
though
they
in
in
their
analysis,
they're,
actually
creating
and
producing
so
there's,
you
know
there's
a
little
bit
of
overlap,
which
is
which
is
okay,
but
in
general,
this
kind
of
this
paradigm
really
worked
for
us
and
then
obviously
we're
down
the
bottom
just
providing
tools
and
trying
to
understand
our
different
users
and
what
they're
trying
to
do
and
thinking
okay
well,
could
we
could
we
give
them
a
service
or
a
tool
that
would
do
that
and
would
make
things
easier
for
them
as
well
as
just
making
you
know,
data
governance
and
data
management,
and
just
a
healthy
data
landscape
in
general
was
is
also
part
of
our
responsibility.
A
So,
along
with
that
sort
of
philosophical
change,
tooling
also
changed
and
and
just
sort
of,
I
guess-
a
real
shift
in
data
becoming
much
more
accessible
by
nature
of
the
way
some
of
these
cloud
computing
technologies
were
going
snowflake.
You
know
the
fact
that
we
we
started
investing
our
time
and
energy
into
building
out
a
snowflake
environment
started
delivering
results
very
very
quickly
for
us
in
terms
of
access
and
just
people,
understanding
and
being
more
confident
with
data.
A
Obviously
dbt
and
airflow,
as
a
main
transformation
stack
and
trying
to
you,
know
letting
best
practice.
Data
and
analytic
and
data
transformation
tooling
become
much
more
accessible
for
people.
A
Obviously,
different
kinds
of
ingestion,
tooling,
like
5tran,
was
also
very
popular
and
tools
like
catalogues
and
lineage
tooling,
like
data
hub,
this
became
has
has
become,
I
think,
more
of
a
secret
weapon
for
us
going
forward
into
data
mesh,
because,
as
we
build
out
the
tooling
for
people
to
work
with
data
mesh,
we
need
to
think
about.
Okay.
Are
they
empowered
to
really
understand
what
this
means?
Do
they
really
get?
What
this
means
and
having
a
tool
to
bring
it
all
to
light?
A
And
make
it
very
user-friendly
to
sort
of
explore
data
is
another
part
of
that
story,
and
so
we're
finding
that
when
we
talk
about
yeah
data
and
metadata
and
and
lineage,
and
things
like
that
really
does
drive
home
to
a
lot
of
our
producers
and
consumers.
What
we
see
every
day,
which
is
how
does
data
flow
across
the
organization
and
can
we
make
it
real
for
them?
A
One
of
the
ways
that
we
it's
improved
is
general
data
literacy
before
there
was
no
easy
way
that
our
consumers
or
producers
could
really
see
definitions,
tags
owners.
Aside
from
looking
at
the
data
contracts,
we
actually
did
collect
all
of
it.
We
collected
it
quite
well
in
the
data
contracting
that
we
had.
A
We
just
didn't
surface
it
particularly
well.
We
didn't
make
it
really
easy
to
find
so
now
we've
got
data
hub,
sucking
in
all
of
that
really
fantastic
metadata
and
just
making
it
really
searchable
in
an
easily
in
an
easy
to
explore.
Ui
before
we
talk
about
governance
and
ownership
responsibilities,
people
kind
of
didn't
really
know
what
that
meant.
It
was
difficult
to
really
know
how
fresh
a
piece
of
data
was
or
who
owns
it
and
who
should
be
responsible
for
keeping
it
updated
for
what
calculated
fields,
because
it
just
wasn't
again
real
to
people.
A
It
wasn't
something
that
they
saw
every
day,
but
now
we
can
use
a
metadata
tool
like
like
data
hub
to
build
workflows,
rank
freshness
and,
just
in
general,
shine
a
light
on
the
distributed
ownership
model
with
lineage,
and
things
like
that.
We
can
easily
see
how
two
different
data
sets
that
might
on
the
surface,
seem
quite
unrelated,
are
related
and
have
been
made
together
and
that's
incredibly
powerful
for
our
community
to
understand
that
they're
part
of
a
of
a
data
community
and
they're
not
just
working
on
their
own
and
they
have
responsibilities
to
their
colleagues.
A
A
Anyone
anytime
anyone
had
a
question
about
data,
they
would
come
to
the
dna
platform
team
and
you
know
half
the
time
we
wouldn't
know
it.
Wasn't
our
data
we'd
know
how
the
tooling
and
the
platform
worked,
but
if
you
had
a
particular
question
about
a
particular
field
or
a
particular
column
and
table
and
why
it
was
being
used,
and
why
is
that
number
look
like
that?
A
lot
of
our
engineers
and
developers
honestly
wouldn't
know
they're,
not
the
subject
matter.
A
Experts
in
that
particular
data
set
now
it's
very
very
easy
for
producers
and
consumers
to
identify
each
other
and
to
be
able
to
directly
interact
with
each
other.
Using
you
know
our
pathways
through
to
email,
psych
or
whatever,
and
just
it
just
makes
it
all
more
efficient.
A
When
you
have
those
kinds
of
questions
that
frees
up
our
time,
it
frees
up
their
time,
hopefully
they're
having
much
more
robust,
interesting
and
fruitful
conversations
than
they
would
have
with
us
anyway,
if
they're
going
directly
to
the
source,
and
all
of
these
things
really
add
to
that
idea
that
we're
all
part
of
this
data
culture
together
and
which
has
been
really
quite
to
be
honest,
I
think
the
key
to
getting
data
mesh
right.
A
How
have
I
gone
for
time?
I'm
not
actually
sure,
but
hopefully
I've
kept.
It
relatively
under
10
minutes.
I'd
really
like
to
thank
the
organizers
for
inviting
me.
I
hope
you
found
something
interesting
to
our
story
and,
if
you're
ever
interested
in
having
a
chat,
please
email
me:
I
am
in
australia,
so
we
might
have
to
do
it
async,
but
I'm
happy
to
chat
and
again,
thank
you
for
having
me.