►
Description
Speaker: Armstrong Foundjem
Organization: Polytechnique Montréal
Software ecosystems bring value by integrating projects related to a given domain, for example, open source projects in a Linux distribution or mobile apps in the Android platform. Since each project has its release cycle and roadmap, having to install compatible versions of each project manually puts an enormous burden on the users, and hence many ecosystems release a polished, well-integrated product to the end user. This paper empirically studies this release synchronization strategy in the context of the OpenStack ecosystem, in which a central release management team manages the six-month release cycle of the overall OpenStack product. By analyzing one year of release team IRC meeting logs, we identified nine major release synchronization activities, which we cataloged and documented. To validate our findings, we interviewed 8 active OpenStack practitioners (members of either the release team or project teams). Our results suggest that even though an ecosystem’s power lies in the interaction of independent projects, release synchronization is a non-trivial goal.
A
Okay,
so
hello,
everyone,
my
name,
is
Armstrong
Hamza
I'm
walking
with
the
M,
the
mining
and
construction
of
intelligence
of
software
laboratory
at
Ecole
Polytechnique
in
Montreal.
So
I
want
to
look
on
the
onboarding
experience,
how
that
actually
plays
a
major
role
in
ecosystem
health.
So,
to
begin
with,
we
try
to
look
at
the
software
ecosystem
from
the
natural
ecosystem
perspective,
to
see
how
much
we
could
learn
from
in
order
to
impact
our
a
software
ecosystem.
So
we
look
at
the
definition
or
the
description
of
sort
of
natural
ecology.
A
We
saw
that
there
are
three
key
words
which
truly
relates
to
software
ecosystem,
unders
words
at
the
interaction
of
organisms
and
their
environments.
So
if
we
look
at
in
the
pictures
on
the
board,
we
see
that
in
the
natural
ecosystem
there
are
several
kinds
of
organisms
that
are
interacting
without
within
the
environment
and
if
we
go
further
now
we
see,
for
example,
there
is
an
organism
that
is
circle
like
the
would
be
fungi.
A
It
has
to
live
on
a
rotten
wood
and
this
sections
inhabited
like
a
home
for
that
sunlight
to
survive
and
that
rotten
wood
cannot
exist
if
the
tree
is
not
fallen.
So
you
see
that
abuse
a
community
which
is
like
the
level
of
a
software
project
where
that
project
is
only
a
subset
of
all
the
projects
that
belong
to
an
ecosystem
and
projects
are
different.
So
what
is
this
organism?
Is
that
this,
this
biological
adaptation
and
interaction
with
the
environment?
A
How
does
the
physical
structure
of
that
organism
interact
with
the
environment
is
very
important
for
us
to
know
how
to
do
an
ecosystem?
Moreover,
we
realized
that
in
the
natural
ecosystem
there
is
this
one
of
bow
diversity,
a
lot
of
people,
especially
the
previous
peak
aspect,
about
diversity.
We
know
the
importance
of
diversity
in
software
ecosystem
software
project
level.
We
need
projects
to
the
different
skills
and
people
to
be
different.
A
Demographic,
our
cultural
and
all
sort
of
diversity
is
highly
working
because
he
enriched
the
project
or
an
ecosystem,
but
we
also
realized
that
in
the
natural
ecosystem,
there
is
this
concept
of
key
organism.
As
we
will
see
later,
when
we
take
the
natural
ecosystem,
we
can
copy-paste
the
concept
into
a
new
software
ecosystem
that
really
fits,
and
those
other
concepts
that
we
have
in
the
natural
ecosystem
can
help
to
boost
the
software
ecosystem.
A
So
when
we
talk
about
the
software
ecosystem,
it
is
built
on
by
people
who
are
enthusiastic
people
who
have
who
are
the
same
driven
skill,
determination,
the
Eider
volunteers
or
the
hired
by
companies
to
work
on
a
project.
Then,
when
you
look
at
an
individual
person
which
happens
to
be
an
organism
in
this
case,
there
are
others.
So
there
are
other
categories
of
organisms
which
we
mention
here
as
a
keystone
organism.
Remember
that
we
boarded
this
concept
also
from
the
natural
ecosystem.
These
are
experts
or
mentors
within
an
ecosystem.
A
They
help
to
train
weaker
organisms,
to
understand
the
culture,
to
understand
the
working
environment,
to
understand
the
code
base
of
a
software
project
so
that
they
could
effectively
contribute,
and
how
does
this
happen?
We
company?
We
came
up
with
this
concept
of
the
niche
which
we
already
mentioned
earlier.
We
see
that
when
you
take
somebody,
for
example,
a
particular
organism,
you
need
to
fit
them
inside
the
projection.
They
need
to
be
at
the
middle,
get
some
form
of
adaptation.
They
need
to
be
adjusted
to
understand
what
is
happening
that
process
itself.
A
If
it
is
not
well
handled,
things
can
even
go
wrong
and
the
better
way
of
handling
this
adaptation,
where
you
fit
people
into
it,
could
be
to
contribute
other
people
who
have
been
there
is
something
that
we
want
to
focus
on
in
this
talk,
this
adaptation
occurs
at
every
step
with
we
investigate
three
different
kinds
of
ditches,
so
this
adaptation
has.
We
set
our
course
now
this,
which
is
how
these
organisms
interact
with
themselves
and
the
environment.
So
is
we
realized
that,
during
a
concept
where
we
were
call
onboarding,
most
of
you
already
know?
A
What
is
one
body
with
there
is
one
particular
niche
is
the
green
area
niche
it
determines
by
organisms
habitat
and
its
behavior
adaptation,
because
people
are
coming
with
trail
knowledge
they
may
have
experienced
with
the
completing,
not
experience,
and
how
do
we
adapt
them
to
that
project?
We
also
see
how
other
form
of
niche
happens,
people
when
they
come
into
a
project
they
need
to
collaborate.
A
There
is
another
kind
of
collaborations
that
we
call
alike
joint
collaboration
when
more
than
one
person's
work
on
a
particular
piece
of
code,
because
it
might
be
so
complex,
it
might
need
more
time.
There
may
be
several
reason
why
the
abilities
joint
collaboration,
this
kind
of
collaboration,
exists
even
in
the
natural
ecosystem,
and
we
have
seen
evidence
in
software
ecosystem.
That
is
also
happening.
So
we
want
to
investigate
this
kind
of
mechanism
how
it
helps
in
the
health
of
an
ecosystem.
A
So
we
also
noticed
that,
when
people
collaborate,
the
end
process
is
either
trees
and
artifacts.
So
this
people
need
to
gain
experience
as
the
a
collaborating
and,
at
the
end,
also
have
to
impact
the
software
ecosystem
or
the
project
they
are
working
with.
Now
this
comes
up.
What
is
actually
the
health
we
are
we
talking
about.
If
we
look
at
from
the
basic
definition
of
what
health
organization
says,
we
are
trying
to
fit
things
together.
A
We
saw
that
the
definition
is
this
type
of
complete
physical,
mental
and
social
well-being,
and
not
just
the
absence
of
disease
or
infirmity.
So
if
we
say
that
a
software
project
is
healthy
because
they
don't
have
book
which
can
be
a
disease
or
they
don't
have
toxicity
which
can
be
infirmity
in
this
case,
we
might
not
be
seeing
a
complete
picture
because
we
might
not
pay
attention
to
the
complete
physical,
the
mental
world.
The
analysis
will
lead
to
ecosystem
health.
A
So
if
we
talk
about
the
complete,
a
complete
physical
as
the
project
must
be
must
be
like
independent
projects
that
are
well
structured.
We
have
team
members
that
knows
what
they
are
doing,
that
mental
capability
did
the
technical
know-how
and
the
social
interactions
that
exist
between
them.
These
are
three
components
that
must
exists,
then
we
can
now
measure
also.
It
is
not
too
much
the
buggy
toxicity
and
the
community
could
cement.
A
If
a
particular
project
is
healthy,
we
cannot
assume
that
the
entire
ecosystem
is
healthy,
so
a
project
help
does
not
equate
to
an
ecosystem
health
at
that
level
of
ecosystem.
But
now
how
do
we
really
see
or
describe
an
ecosystem
here?
From
this
perspective,
we
see
that
if
everything,
if
every
project
team
and
every
key
player
is
operating
properly,
we
can
now
say
that
the
entire
ecosystem
is
sound
is
healthy.
So
we
did
a
case
study
with
OpenStack
ecosystem,
which
many
of
the
previous
speaker
also
mentioned.
I
said
it
is
a
successful
ecosystem.
A
A
So
we
look
at
the
elephant
factor
which
others
have
already
mentioned,
the
clip
of
Condor
and
things
like
that.
We
found
them
in
different
these
four
categories.
We're
talking
about
all
these
things.
We
realized
that
the
quest
community
has
already
set
up
a
very
good
baseline,
whereby
we
can
easily
measure
all
those
technical
issues
to
describe
the
health
not
only
of
a
project
bit
of
an
ecosystem
so
to
go
into
the
context
of
our
talk
today.
A
We
try
to
look
at
on
body
how
it
impacts
the
health
of
an
ecosystem
that,
by
the
way,
what
is
one
body
just
a
definition
that
the
dictionary
gives
us
for
own
body.
This
is
an
action
or
process
of
integrating
new
employees
into
an
a
goes
into
an
organization
or
familiarizing
new
customers
or
client
with
the
process
of
a
services.
But
we
realize
that
the
symbol
of
scissors
that
we
are
trying
to
cut
out
that
word
action
because
of
our
observation
and
understanding
that
weaves
it
one
body
is
actually
a
process
and
not
an
action.
A
We
will
show
that
later
on.
Why
so,
with
the
case
of
OpenStack
ecosystem,
which
is
a
well-structured
ecosystem,
they
have
this
concept
of
one
body
and
it
has
been
proven
to
be
successful
for
our
ecosystem.
How
can
we
not
take
this
experience
to
generalize
it
to
any
other
ecosystem?
So
we
realized
we
did
an
observational
study.
We
saw
the
the
diversity
of
people
and
skills
that
are
covering
for
the
unburdening
and
how
they
are
training,
new
people
to
be
effective
within
that
ecosystem.
A
So
OpenStack
actually
introduced
own
body
in
2014,
but
before
OpenStack
itself
was
created
in
2010
from
Rackspace
and
NASA
that
we
all
know
now
we
try
to
measure
the
quality
of
contributors.
We
use
effort
the
time
that
they
took
to
do
DFS
accepted
common
and
many
other
matrix
is
from
2014
that
OpenStack
was
introduced.
We
separated
two
categories
of
people
like
those
who
did
the
own
body
and
those
who
did
not
do
it
to
see
how
that
impacts.
If
one
body
actually
has
an
impact,
the
ecosystem,
health.
A
So
based
on
that,
we
realized
that
those
who
did
own
body-
we
saw
before
sorry
the
performance
of
this
before
own
body,
and
we
did
not
do
unberthing.
There
was
a
great
significant
difference
towards
the
efest
commit
those
who
did
own
body
actually
spend
a
lesser
time
to
do
DFS
acceptable
commit
than
those
who
did
not
do
they
had
to
do
several
iterations
before
it
was
accepted.
A
A
Which
slices
that
you
wonder
for
this
sorry?
This
is
the
survivor
analysis.
In
this
survivor
analysis.
We
are
trying
to
measure
it
time
to
an
event
a
time
before
an
event
or
call
and
in
the
x
axis
is
the
time
is
measured
by
the
number
of
these.
So
we
want
to
see
the
monthly
the
the
number
of
days
in
which
the
contributors
actually
did.
The
efest
called
submitted
the
efest
part
and
it
was
accepted.
So
if
we
see
the
colors
that
we
use
there,
we
try
to
separate
the
contributors
pair
the
gender
diversity.
A
So
we
look
at
people
who
have
like
male
female
and
the
Nutri
category.
We
saw
how
that
diversity
of
the
group
influences.
We
like
their
contribution
in
terms
of
the
effort
that
we
measure
and
the
commit
that
they
were
submitting.
So
we
are
trying
to
look
also
the
diversity
in
that
ecosystem
and
now
to
see
how
each
category
is
performing
over
time.
A
A
So
as
time
goes
on,
those
who
did
one
body
have
the
efest
acceptable
commit
faster
than
those
who
did
not
do
on
body.
So
that
is
an
indication
that
one
body
helps
to
build
a
healthy
ecosystem.
You
know,
because
we
talked
about
accepted,
they
have
gone
through
all
they
could
reviews
and
things
before
it
is
accepted.
So
we
cannot
measure
from
there.
We
cannot
start
measuring
qualities
and
things
like
that
when
we
say
we
look
at
the
effort
like
here
where
we're
measuring
the
number
of
attempts.
A
That
is
it
all
verdict
at
the
attract
committee
itself,
so
the
longer
it
takes
they
make
more.
Attempts
means
there
are
things
to
improve.
There
are
things
to
improve
before
it
accident
final
acceptance.
So
if
we
realize
that
the
shorter
the
effort,
the
better
and
the
higher
quality
it
is
so
this
would
no
equality
described.
Those
would
higher
quality
with
sorry.
A
So
sorry
that
was
firstly
law,
so
just
to
combat
okay,
we
realize
that
mentoring,
as
we
saw,
was
one
active.
We
were
useful
instrument
which
this
one
body
introduced
like
in
OpenStack.
What
we
observe
is
that
when
people
come
in
for
the
onboarding
training,
they
are
also
mentored
throughout
the
process.
It
is
not
just
an
event
that
happen,
and
it
is,
it
is
done.
They
try
to
fold
them
up,
send
remind
that
sometimes
get
one-on-one
communication.
We
noticed
that
we
attempted
this.
Do
you
have?
A
Do
you
have
difficulties
in
completing
this
kind
of
show
up
helps
a
lot
of
people,
as
we
observed
in
our
findings
later
on
to
the
active
within
the
community
and
afterward
to
remain
within
the
community.
So
this
also
shows
us.
Some
indications
that,
under
you
is
not
just
like
unloading,
is
not
an
event
that
a
process
that
fools
of
people
up
to
an
integration
and
that
can
also
show
indication
to
reduce
abandoning,
because
if
you
understand
that
we
can
mentor,
people
understand
that
at
some
point
they
are
not
getting
involved
anymore.
A
It's
already
it
and
get
something
is
wrong.
Then
full
up
can
take
place
from
there
so
to
completely
introduce
this
concept
of
own
body
model
from
the
natural
ecosystem,
which
already
is
something
that
human
being
can
is
already
involved
and
can
learn
from.
We
realize
that
we
can
copy
this
concept.
We
can
borrow
it
from
the
natural
ecosystem
to
software
ecosystem.
I
fits
very
well.
A
We
also
introduce
one
body
which
is
a
tool
to
measure
the
software
ecosystem,
health,
and
we
saw
that
onboarding
was
actually
a
process
and
not
an
action
or
just
an
event
that
happens,
and
it
is
done
so.
We
introduced
some
matrixes
where
which
we
use
to
show
that
OpenStack,
for
example,
is
a
well
diverse
and
inclusive
ecosystem.
A
It
has
this
evolution
and
the
survival
of
the
contributors
over
time
it
has
grown
to
cross
community
collaboration
where
you
can
see
different
projects
working
together
to
build
a
healthy
ecosystem
and
also
we
show
now
the
survival
analysis
which
shows
that
those
who
deep
on
body
performs
better
over
time
than
those
who
did
not
do,
even
though
there
are
a
lot
of
our
explanation
that
can
still
the
given.
We
have
taken
care
to
that
and
we,
we
are
still
doing
a
studies
to
to
come
up
with
a
bigger
conclusion.
B
B
A
Yeah
we
actually
before
doing
the
this
study.
We
consider
communities
a
lot
of
other
communities,
for
example
Apache
ecosystem,
the
Eclipse
and
many
others
that,
because
of
the
nature
of
our
study,
this
is
just
one
aspect
of
it.
We
needed
an
ecosystem
that
is
worth
structure
to
begin
with,
because
we
need
to
measure
the
health,
we
need
something
that
we
can
measure.
So
that
is
one
of
the
reason
equal,
like
OpenStack
was
really
a
favorable
candidate.
A
Then
we
actually
move
into
an
extent
Apache
and
the
Eclipse
and
many
others,
and
realize
that
some
of
them
were
lacking
in
some
other
criterias
that
we
needed
to
be
in
place.
But
it
also
be
a
good
thing
at
the
end
of
the
day,
to
see
if
our
recommendations
and
things
like
that
become
pupas,
where
we
can
actually
suggest
up
to
such
others
of
those
ecosystems.
A
The
point
there
with
a
case
like
Red
Hat,
which
is
already
an
established
project
and
is
building
itself
also
to
an
ecosystem.
They
think
in
with
onboarding
here,
will
be
when
people
come
in.
How
do
they
understand,
for
example,
the
ecosystem
itself,
and
where
can
they
actually
contribute
because
new
members,
when
they
come
into
a
project,
sometimes
they
need
someone
to
navigate
them
to
show
with
a
vivid
overview
of
the
project,
let
alone
some
of
them
might
discover
it
with
some
can
be
easily
get
lost
there
at
time
to
discover
own
body.
A
Try
to
give
a
focus
so
in
that
context,
on
body
will
have
to
give
a
focus.
It's
not
like
giving
the
answer
to
the
contributor
that
this
is
what
we
should
do,
but
guiding
them
towards
their
skills.
We
have
seen
cases
where
also
people
started
in
one
project,
because
why
everybody
is
navigating
towards
that
direction
only
after
a
period
of
six
months.
Oh
no!
A
C
A
The
case
of
open
stab
what
I
observe
personal
is
that
when
there
is
an
op,
they
have
what
is
called
an
upstream
Institute,
a
particular
team
of
experienced
people
that
actually
handles
on
body,
and
this
expert
comes
from
cords
developers
from
other
projects
within
the
eCos
in
OpenStack
to
help
the
younger
people
to
grow,
because
we
know
there
are
so
many
projects
within
OpenStack
and
people
coming
in.
They
need
to
know
they
are
way
out.
So
this
core
developers
from
other
projects
come
Nvidia
skills
to
help
these
new
people.
A
So
during
the
onboarding
events,
these
experts,
they
organize
these
events.
The
company
are
there
physically,
and
these
new
people
are
there.
Then
they
start
following
them
up
even
after
the
to
this
event
is
over,
they
still
have
a
means
of
communication
of
communicating
with
these
people
and
to
the
best
of
my
knowledge,
I
know
if
you
still
go
and
sign
up
with
any
of
the
IRC
channel
ask
for
help,
even
if
you
do
not
come
for
onboarding
and
show
some
grip,
enthusiastic
skills
like
seriousness,
though
always
with
people
to
help
you
around.