<s>
A	O
distributed	B-Architecture
system	I-Architecture
is	O
a	O
system	O
whose	O
components	O
are	O
located	O
on	O
different	O
networked	B-Architecture
computers	I-Architecture
,	O
which	O
communicate	O
and	O
coordinate	O
their	O
actions	O
by	O
passing	B-Architecture
messages	I-Architecture
to	O
one	O
another	O
.	O
</s>
<s>
Distributed	B-Architecture
computing	I-Architecture
is	O
a	O
field	O
of	O
computer	B-General_Concept
science	I-General_Concept
that	O
studies	O
distributed	B-Architecture
systems	I-Architecture
.	O
</s>
<s>
The	O
components	O
of	O
a	O
distributed	B-Architecture
system	I-Architecture
interact	O
with	O
one	O
another	O
in	O
order	O
to	O
achieve	O
a	O
common	O
goal	O
.	O
</s>
<s>
Three	O
significant	O
challenges	O
of	O
distributed	B-Architecture
systems	I-Architecture
are	O
:	O
maintaining	O
concurrency	B-Application
of	O
components	O
,	O
overcoming	O
the	O
lack	O
of	O
a	O
global	O
clock	O
,	O
and	O
managing	O
the	O
independent	O
failure	O
of	O
components	O
.	O
</s>
<s>
Examples	O
of	O
distributed	B-Architecture
systems	I-Architecture
vary	O
from	O
SOA-based	O
systems	O
to	O
massively	B-Application
multiplayer	I-Application
online	I-Application
games	I-Application
to	O
peer-to-peer	O
applications	O
.	O
</s>
<s>
A	O
computer	B-Application
program	I-Application
that	O
runs	O
within	O
a	O
distributed	B-Architecture
system	I-Architecture
is	O
called	O
a	O
distributed	B-Architecture
program	I-Architecture
,	O
and	O
distributed	B-Architecture
programming	I-Architecture
is	O
the	O
process	B-Operating_System
of	O
writing	O
such	O
programs	O
.	O
</s>
<s>
There	O
are	O
many	O
different	O
types	O
of	O
implementations	O
for	O
the	O
message	B-Architecture
passing	I-Architecture
mechanism	O
,	O
including	O
pure	O
HTTP	O
,	O
RPC-like	B-Operating_System
connectors	O
and	O
message	B-Application
queues	I-Application
.	O
</s>
<s>
Distributed	B-Architecture
computing	I-Architecture
also	O
refers	O
to	O
the	O
use	O
of	O
distributed	B-Architecture
systems	I-Architecture
to	O
solve	O
computational	O
problems	O
.	O
</s>
<s>
In	O
distributed	B-Architecture
computing	I-Architecture
,	O
a	O
problem	O
is	O
divided	O
into	O
many	O
tasks	O
,	O
each	O
of	O
which	O
is	O
solved	O
by	O
one	O
or	O
more	O
computers	O
,	O
which	O
communicate	O
with	O
each	O
other	O
via	O
message	B-Architecture
passing	I-Architecture
.	O
</s>
<s>
The	O
word	O
distributed	O
in	O
terms	O
such	O
as	O
"	O
distributed	B-Architecture
system	I-Architecture
"	O
,	O
"	O
distributed	B-Architecture
programming	I-Architecture
"	O
,	O
and	O
"	O
distributed	B-Operating_System
algorithm	I-Operating_System
"	O
originally	O
referred	O
to	O
computer	B-Architecture
networks	I-Architecture
where	O
individual	O
computers	O
were	O
physically	O
distributed	O
within	O
some	O
geographical	O
area	O
.	O
</s>
<s>
The	O
terms	O
are	O
nowadays	O
used	O
in	O
a	O
much	O
wider	O
sense	O
,	O
even	O
referring	O
to	O
autonomous	O
processes	B-Operating_System
that	O
run	O
on	O
the	O
same	O
physical	O
computer	O
and	O
interact	O
with	O
each	O
other	O
by	O
message	B-Architecture
passing	I-Architecture
.	O
</s>
<s>
While	O
there	O
is	O
no	O
single	O
definition	O
of	O
a	O
distributed	B-Architecture
system	I-Architecture
,	O
the	O
following	O
defining	O
properties	O
are	O
commonly	O
used	O
as	O
:	O
</s>
<s>
There	O
are	O
several	O
autonomous	O
computational	O
entities	O
(	O
computers	O
or	O
nodes	B-Protocol
)	O
,	O
each	O
of	O
which	O
has	O
its	O
own	O
local	O
memory	B-General_Concept
.	O
</s>
<s>
The	O
entities	O
communicate	O
with	O
each	O
other	O
by	O
message	B-Architecture
passing	I-Architecture
.	O
</s>
<s>
A	O
distributed	B-Architecture
system	I-Architecture
may	O
have	O
a	O
common	O
goal	O
,	O
such	O
as	O
solving	O
a	O
large	O
computational	O
problem	O
;	O
the	O
user	O
then	O
perceives	O
the	O
collection	O
of	O
autonomous	O
processors	O
as	O
a	O
unit	O
.	O
</s>
<s>
Alternatively	O
,	O
each	O
computer	O
may	O
have	O
its	O
own	O
user	O
with	O
individual	O
needs	O
,	O
and	O
the	O
purpose	O
of	O
the	O
distributed	B-Architecture
system	I-Architecture
is	O
to	O
coordinate	O
the	O
use	O
of	O
shared	O
resources	O
or	O
provide	O
communication	O
services	O
to	O
the	O
users	O
.	O
</s>
<s>
Other	O
typical	O
properties	O
of	O
distributed	B-Architecture
systems	I-Architecture
include	O
the	O
following	O
:	O
</s>
<s>
The	O
system	O
has	O
to	O
tolerate	B-General_Concept
failures	I-General_Concept
in	O
individual	O
computers	O
.	O
</s>
<s>
The	O
structure	O
of	O
the	O
system	O
(	O
network	O
topology	O
,	O
network	O
latency	O
,	O
number	O
of	O
computers	O
)	O
is	O
not	O
known	O
in	O
advance	O
,	O
the	O
system	O
may	O
consist	O
of	O
different	O
kinds	O
of	O
computers	O
and	O
network	O
links	O
,	O
and	O
the	O
system	O
may	O
change	O
during	O
the	O
execution	O
of	O
a	O
distributed	B-Architecture
program	I-Architecture
.	O
</s>
<s>
Distributed	B-Architecture
systems	I-Architecture
are	O
groups	O
of	O
networked	B-Architecture
computers	I-Architecture
which	O
share	O
a	O
common	O
goal	O
for	O
their	O
work	O
.	O
</s>
<s>
The	O
terms	O
"	O
concurrent	B-Architecture
computing	I-Architecture
"	O
,	O
"	O
parallel	B-Operating_System
computing	I-Operating_System
"	O
,	O
and	O
"	O
distributed	B-Architecture
computing	I-Architecture
"	O
have	O
much	O
overlap	O
,	O
and	O
no	O
clear	O
distinction	O
exists	O
between	O
them	O
.	O
</s>
<s>
The	O
same	O
system	O
may	O
be	O
characterized	O
both	O
as	O
"	O
parallel	O
"	O
and	O
"	O
distributed	O
"	O
;	O
the	O
processors	O
in	O
a	O
typical	O
distributed	B-Architecture
system	I-Architecture
run	O
concurrently	O
in	O
parallel	O
.	O
</s>
<s>
Parallel	B-Operating_System
computing	I-Operating_System
may	O
be	O
seen	O
as	O
a	O
particular	O
tightly	O
coupled	O
form	O
of	O
distributed	B-Architecture
computing	I-Architecture
,	O
and	O
distributed	B-Architecture
computing	I-Architecture
may	O
be	O
seen	O
as	O
a	O
loosely	B-Application
coupled	I-Application
form	O
of	O
parallel	B-Operating_System
computing	I-Operating_System
.	O
</s>
<s>
Nevertheless	O
,	O
it	O
is	O
possible	O
to	O
roughly	O
classify	O
concurrent	B-Operating_System
systems	O
as	O
"	O
parallel	O
"	O
or	O
"	O
distributed	O
"	O
using	O
the	O
following	O
criteria	O
:	O
</s>
<s>
In	O
parallel	B-Operating_System
computing	I-Operating_System
,	O
all	O
processors	O
may	O
have	O
access	O
to	O
a	O
shared	B-Operating_System
memory	I-Operating_System
to	O
exchange	O
information	O
between	O
processors	O
.	O
</s>
<s>
In	O
distributed	B-Architecture
computing	I-Architecture
,	O
each	O
processor	O
has	O
its	O
own	O
private	O
memory	B-General_Concept
(	O
distributed	B-Operating_System
memory	I-Operating_System
)	O
.	O
</s>
<s>
Information	O
is	O
exchanged	O
by	O
passing	B-Architecture
messages	I-Architecture
between	O
the	O
processors	O
.	O
</s>
<s>
Figure	O
(	O
a	O
)	O
is	O
a	O
schematic	O
view	O
of	O
a	O
typical	O
distributed	B-Architecture
system	I-Architecture
;	O
the	O
system	O
is	O
represented	O
as	O
a	O
network	O
topology	O
in	O
which	O
each	O
node	O
is	O
a	O
computer	O
and	O
each	O
line	O
connecting	O
the	O
nodes	B-Protocol
is	O
a	O
communication	O
link	O
.	O
</s>
<s>
Figure	O
(	O
b	O
)	O
shows	O
the	O
same	O
distributed	B-Architecture
system	I-Architecture
in	O
more	O
detail	O
:	O
each	O
computer	O
has	O
its	O
own	O
local	O
memory	B-General_Concept
,	O
and	O
information	O
can	O
be	O
exchanged	O
only	O
by	O
passing	B-Architecture
messages	I-Architecture
from	O
one	O
node	O
to	O
another	O
by	O
using	O
the	O
available	O
communication	O
links	O
.	O
</s>
<s>
Figure	O
(	O
c	O
)	O
shows	O
a	O
parallel	O
system	O
in	O
which	O
each	O
processor	O
has	O
a	O
direct	O
access	O
to	O
a	O
shared	B-Operating_System
memory	I-Operating_System
.	O
</s>
<s>
The	O
situation	O
is	O
further	O
complicated	O
by	O
the	O
traditional	O
uses	O
of	O
the	O
terms	O
parallel	O
and	O
distributed	B-Operating_System
algorithm	I-Operating_System
that	O
do	O
not	O
quite	O
match	O
the	O
above	O
definitions	O
of	O
parallel	O
and	O
distributed	B-Architecture
systems	I-Architecture
(	O
see	O
below	O
for	O
more	O
detailed	O
discussion	O
)	O
.	O
</s>
<s>
Nevertheless	O
,	O
as	O
a	O
rule	O
of	O
thumb	O
,	O
high-performance	O
parallel	B-Operating_System
computation	I-Operating_System
in	O
a	O
shared-memory	B-Operating_System
multiprocessor	I-Operating_System
uses	O
parallel	O
algorithms	O
while	O
the	O
coordination	O
of	O
a	O
large-scale	O
distributed	B-Architecture
system	I-Architecture
uses	O
distributed	B-Operating_System
algorithms	I-Operating_System
.	O
</s>
<s>
The	O
use	O
of	O
concurrent	B-Operating_System
processes	B-Operating_System
which	O
communicate	O
through	O
message-passing	B-Architecture
has	O
its	O
roots	O
in	O
operating	B-General_Concept
system	I-General_Concept
architectures	O
studied	O
in	O
the	O
1960s	O
.	O
</s>
<s>
The	O
first	O
widespread	O
distributed	B-Architecture
systems	I-Architecture
were	O
local-area	B-General_Concept
networks	I-General_Concept
such	O
as	O
Ethernet	O
,	O
which	O
was	O
invented	O
in	O
the	O
1970s	O
.	O
</s>
<s>
E-mail	O
became	O
the	O
most	O
successful	O
application	O
of	O
ARPANET	O
,	O
and	O
it	O
is	O
probably	O
the	O
earliest	O
example	O
of	O
a	O
large-scale	O
distributed	B-Architecture
application	I-Architecture
.	O
</s>
<s>
In	O
addition	O
to	O
ARPANET	O
(	O
and	O
its	O
successor	O
,	O
the	O
global	O
Internet	O
)	O
,	O
other	O
early	O
worldwide	O
computer	B-Architecture
networks	I-Architecture
included	O
Usenet	B-Application
and	O
FidoNet	O
from	O
the	O
1980s	O
,	O
both	O
of	O
which	O
were	O
used	O
to	O
support	O
distributed	O
discussion	O
systems	O
.	O
</s>
<s>
The	O
study	O
of	O
distributed	B-Architecture
computing	I-Architecture
became	O
its	O
own	O
branch	O
of	O
computer	B-General_Concept
science	I-General_Concept
in	O
the	O
late	O
1970s	O
and	O
early	O
1980s	O
.	O
</s>
<s>
The	O
first	O
conference	O
in	O
the	O
field	O
,	O
Symposium	O
on	O
Principles	O
of	O
Distributed	B-Architecture
Computing	I-Architecture
(	O
PODC	O
)	O
,	O
dates	O
back	O
to	O
1982	O
,	O
and	O
its	O
counterpart	O
International	O
Symposium	O
on	O
Distributed	B-Architecture
Computing	I-Architecture
(	O
DISC	O
)	O
was	O
first	O
held	O
in	O
Ottawa	O
in	O
1985	O
as	O
the	O
International	O
Workshop	O
on	O
Distributed	B-Operating_System
Algorithms	I-Operating_System
on	O
Graphs	O
.	O
</s>
<s>
Various	O
hardware	O
and	O
software	O
architectures	O
are	O
used	O
for	O
distributed	B-Architecture
computing	I-Architecture
.	O
</s>
<s>
At	O
a	O
lower	O
level	O
,	O
it	O
is	O
necessary	O
to	O
interconnect	O
multiple	O
CPUs	O
with	O
some	O
sort	O
of	O
network	O
,	O
regardless	O
of	O
whether	O
that	O
network	O
is	O
printed	O
onto	O
a	O
circuit	O
board	O
or	O
made	O
up	O
of	O
loosely	B-Application
coupled	I-Application
devices	O
and	O
cables	O
.	O
</s>
<s>
At	O
a	O
higher	O
level	O
,	O
it	O
is	O
necessary	O
to	O
interconnect	O
processes	B-Operating_System
running	O
on	O
those	O
CPUs	O
with	O
some	O
sort	O
of	O
communication	O
system	O
.	O
</s>
<s>
Distributed	B-Architecture
programming	I-Architecture
typically	O
falls	O
into	O
one	O
of	O
several	O
basic	O
architectures	O
:	O
client	B-Operating_System
–	I-Operating_System
server	I-Operating_System
,	O
three-tier	B-Operating_System
,	O
n-tier	B-Operating_System
,	O
or	O
peer-to-peer	O
;	O
or	O
categories	O
:	O
loose	B-Application
coupling	I-Application
,	O
or	O
tight	B-Architecture
coupling	I-Architecture
.	O
</s>
<s>
Client	B-Operating_System
–	I-Operating_System
server	I-Operating_System
:	O
architectures	O
where	O
smart	O
clients	O
contact	O
the	O
server	O
for	O
data	O
then	O
format	O
and	O
display	O
it	O
to	O
the	O
users	O
.	O
</s>
<s>
Three-tier	B-Operating_System
:	O
architectures	O
that	O
move	O
the	O
client	O
intelligence	O
to	O
a	O
middle	B-Operating_System
tier	I-Operating_System
so	O
that	O
stateless	B-Protocol
clients	O
can	O
be	O
used	O
.	O
</s>
<s>
Most	O
web	O
applications	O
are	O
three-tier	B-Operating_System
.	O
</s>
<s>
n-tier	B-Operating_System
:	O
architectures	O
that	O
refer	O
typically	O
to	O
web	O
applications	O
which	O
further	O
forward	O
their	O
requests	O
to	O
other	O
enterprise	O
services	O
.	O
</s>
<s>
This	O
type	O
of	O
application	O
is	O
the	O
one	O
most	O
responsible	O
for	O
the	O
success	O
of	O
application	B-Application
servers	I-Application
.	O
</s>
<s>
Examples	O
of	O
this	O
architecture	O
include	O
BitTorrent	B-Protocol
and	O
the	O
bitcoin	B-General_Concept
network	I-General_Concept
.	O
</s>
<s>
Another	O
basic	O
aspect	O
of	O
distributed	B-Architecture
computing	I-Architecture
architecture	I-Architecture
is	O
the	O
method	O
of	O
communicating	O
and	O
coordinating	O
work	O
among	O
concurrent	B-Operating_System
processes	B-Operating_System
.	O
</s>
<s>
Through	O
various	O
message	B-Architecture
passing	I-Architecture
protocols	O
,	O
processes	B-Operating_System
may	O
communicate	O
directly	O
with	O
one	O
another	O
,	O
typically	O
in	O
a	O
master/slave	B-Operating_System
relationship	O
.	O
</s>
<s>
Alternatively	O
,	O
a	O
"	B-Operating_System
database-centric	I-Operating_System
"	I-Operating_System
architecture	I-Operating_System
can	O
enable	O
distributed	B-Architecture
computing	I-Architecture
to	O
be	O
done	O
without	O
any	O
form	O
of	O
direct	O
inter-process	B-Operating_System
communication	I-Operating_System
,	O
by	O
utilizing	O
a	O
shared	O
database	O
.	O
</s>
<s>
Database-centric	B-Operating_System
architecture	I-Operating_System
in	O
particular	O
provides	O
relational	O
processing	O
analytics	O
in	O
a	O
schematic	O
architecture	O
allowing	O
for	O
live	O
environment	O
relay	O
.	O
</s>
<s>
This	O
enables	O
distributed	B-Architecture
computing	I-Architecture
functions	O
both	O
within	O
and	O
beyond	O
the	O
parameters	O
of	O
a	O
networked	O
database	O
.	O
</s>
<s>
Reasons	O
for	O
using	O
distributed	B-Architecture
systems	I-Architecture
and	O
distributed	B-Architecture
computing	I-Architecture
may	O
include	O
:	O
</s>
<s>
There	O
are	O
many	O
cases	O
in	O
which	O
the	O
use	O
of	O
a	O
single	O
computer	O
would	O
be	O
possible	O
in	O
principle	O
,	O
but	O
the	O
use	O
of	O
a	O
distributed	B-Architecture
system	I-Architecture
is	O
beneficial	O
for	O
practical	O
reasons	O
.	O
</s>
<s>
It	O
can	O
allow	O
for	O
much	O
larger	O
storage	O
and	O
memory	B-General_Concept
,	O
faster	O
compute	O
,	O
and	O
higher	O
bandwidth	O
than	O
a	O
single	O
machine	O
.	O
</s>
<s>
It	O
can	O
provide	O
more	O
reliability	O
than	O
a	O
non-distributed	O
system	O
,	O
as	O
there	O
is	O
no	B-Architecture
single	I-Architecture
point	I-Architecture
of	I-Architecture
failure	I-Architecture
.	O
</s>
<s>
Moreover	O
,	O
a	O
distributed	B-Architecture
system	I-Architecture
may	O
be	O
easier	O
to	O
expand	O
and	O
manage	O
than	O
a	O
monolithic	O
uniprocessor	O
system	O
.	O
</s>
<s>
It	O
may	O
be	O
more	O
cost-efficient	O
to	O
obtain	O
the	O
desired	O
level	O
of	O
performance	O
by	O
using	O
a	O
cluster	B-Architecture
of	O
several	O
low-end	O
computers	O
,	O
in	O
comparison	O
with	O
a	O
single	O
high-end	O
computer	O
.	O
</s>
<s>
Examples	O
of	O
distributed	B-Architecture
systems	I-Architecture
and	O
applications	O
of	O
distributed	B-Architecture
computing	I-Architecture
include	O
the	O
following	O
:	O
</s>
<s>
computer	B-Architecture
networks	I-Architecture
such	O
as	O
the	O
Internet	O
,	O
</s>
<s>
wireless	B-Architecture
sensor	I-Architecture
networks	I-Architecture
,	O
</s>
<s>
routing	B-Protocol
algorithms	I-Protocol
;	O
</s>
<s>
massively	B-Application
multiplayer	I-Application
online	I-Application
games	I-Application
and	O
virtual	B-Application
reality	I-Application
communities	O
,	O
</s>
<s>
distributed	B-Application
databases	I-Application
and	O
distributed	B-Application
database	I-Application
management	I-Application
systems	I-Application
,	O
</s>
<s>
distributed	B-Operating_System
cache	I-Operating_System
such	O
as	O
burst	B-Operating_System
buffers	I-Operating_System
,	O
</s>
<s>
distributed	B-Architecture
information	I-Architecture
processing	I-Architecture
systems	O
such	O
as	O
banking	O
systems	O
and	O
airline	O
reservation	O
systems	O
;	O
</s>
<s>
real-time	O
process	B-Operating_System
control	O
:	O
</s>
<s>
parallel	B-Operating_System
computation	I-Operating_System
:	O
</s>
<s>
scientific	O
computing	O
,	O
including	O
cluster	B-Architecture
computing	I-Architecture
,	O
grid	B-Architecture
computing	I-Architecture
,	O
cloud	B-Architecture
computing	I-Architecture
,	O
and	O
various	O
volunteer	O
computing	O
projects	O
,	O
</s>
<s>
distributed	B-Operating_System
rendering	I-Operating_System
in	O
computer	O
graphics	O
.	O
</s>
<s>
In	O
theoretical	O
computer	B-General_Concept
science	I-General_Concept
,	O
such	O
tasks	O
are	O
called	O
computational	O
problems	O
.	O
</s>
<s>
Theoretical	O
computer	B-General_Concept
science	I-General_Concept
seeks	O
to	O
understand	O
which	O
computational	O
problems	O
can	O
be	O
solved	O
by	O
using	O
a	O
computer	O
(	O
computability	O
theory	O
)	O
and	O
how	O
efficiently	O
(	O
computational	O
complexity	O
theory	O
)	O
.	O
</s>
<s>
Such	O
an	O
algorithm	O
can	O
be	O
implemented	O
as	O
a	O
computer	B-Application
program	I-Application
that	O
runs	O
on	O
a	O
general-purpose	O
computer	O
:	O
the	O
program	O
reads	O
a	O
problem	O
instance	O
from	O
input	O
,	O
performs	O
some	O
computation	O
,	O
and	O
produces	O
the	O
solution	O
as	O
output	B-General_Concept
.	O
</s>
<s>
Formalisms	O
such	O
as	O
random-access	B-Application
machines	I-Application
or	O
universal	O
Turing	O
machines	O
can	O
be	O
used	O
as	O
abstract	O
models	O
of	O
a	O
sequential	O
general-purpose	O
computer	O
executing	O
such	O
an	O
algorithm	O
.	O
</s>
<s>
The	O
field	O
of	O
concurrent	B-Operating_System
and	O
distributed	B-Architecture
computing	I-Architecture
studies	O
similar	O
questions	O
in	O
the	O
case	O
of	O
either	O
multiple	O
computers	O
,	O
or	O
a	O
computer	O
that	O
executes	O
a	O
network	O
of	O
interacting	O
processes	B-Operating_System
:	O
which	O
computational	O
problems	O
can	O
be	O
solved	O
in	O
such	O
a	O
network	O
and	O
how	O
efficiently	O
?	O
</s>
<s>
However	O
,	O
it	O
is	O
not	O
at	O
all	O
obvious	O
what	O
is	O
meant	O
by	O
"	O
solving	O
a	O
problem	O
"	O
in	O
the	O
case	O
of	O
a	O
concurrent	B-Operating_System
or	O
distributed	B-Architecture
system	I-Architecture
:	O
for	O
example	O
,	O
what	O
is	O
the	O
task	O
of	O
the	O
algorithm	O
designer	O
,	O
and	O
what	O
is	O
the	O
concurrent	B-Operating_System
or	O
distributed	O
equivalent	O
of	O
a	O
sequential	O
general-purpose	O
computer	O
?	O
</s>
<s>
The	O
discussion	O
below	O
focuses	O
on	O
the	O
case	O
of	O
multiple	O
computers	O
,	O
although	O
many	O
of	O
the	O
issues	O
are	O
the	O
same	O
for	O
concurrent	B-Operating_System
processes	B-Operating_System
running	O
on	O
a	O
single	O
computer	O
.	O
</s>
<s>
All	O
processors	O
have	O
access	O
to	O
a	O
shared	B-Operating_System
memory	I-Operating_System
.	O
</s>
<s>
One	O
theoretical	O
model	O
is	O
the	O
parallel	B-Operating_System
random-access	I-Operating_System
machines	I-Operating_System
(	O
PRAM	O
)	O
that	O
are	O
used	O
.	O
</s>
<s>
However	O
,	O
the	O
classical	O
PRAM	O
model	O
assumes	O
synchronous	O
access	O
to	O
the	O
shared	B-Operating_System
memory	I-Operating_System
.	O
</s>
<s>
Shared-memory	B-Operating_System
programs	O
can	O
be	O
extended	O
to	O
distributed	B-Architecture
systems	I-Architecture
if	O
the	O
underlying	O
operating	B-General_Concept
system	I-General_Concept
encapsulates	O
the	O
communication	O
between	O
nodes	B-Protocol
and	O
virtually	O
unifies	O
the	O
memory	B-General_Concept
across	O
all	O
individual	O
systems	O
.	O
</s>
<s>
A	O
model	O
that	O
is	O
closer	O
to	O
the	O
behavior	O
of	O
real-world	O
multiprocessor	O
machines	O
and	O
takes	O
into	O
account	O
the	O
use	O
of	O
machine	O
instructions	O
,	O
such	O
as	O
Compare-and-swap	B-Operating_System
(	O
CAS	O
)	O
,	O
is	O
that	O
of	O
asynchronous	O
shared	B-Operating_System
memory	I-Operating_System
.	O
</s>
<s>
Models	O
such	O
as	O
Boolean	O
circuits	O
and	O
sorting	B-Algorithm
networks	I-Algorithm
are	O
used	O
.	O
</s>
<s>
A	O
Boolean	O
circuit	O
can	O
be	O
seen	O
as	O
a	O
computer	B-Architecture
network	I-Architecture
:	O
each	O
gate	O
is	O
a	O
computer	O
that	O
runs	O
an	O
extremely	O
simple	O
computer	B-Application
program	I-Application
.	O
</s>
<s>
Similarly	O
,	O
a	O
sorting	B-Algorithm
network	I-Algorithm
can	O
be	O
seen	O
as	O
a	O
computer	B-Architecture
network	I-Architecture
:	O
each	O
comparator	O
is	O
a	O
computer	O
.	O
</s>
<s>
The	O
algorithm	O
designer	O
only	O
chooses	O
the	O
computer	B-Application
program	I-Application
.	O
</s>
<s>
A	O
commonly	O
used	O
model	O
is	O
a	O
graph	O
with	O
one	O
finite-state	B-Architecture
machine	I-Architecture
per	O
node	O
.	O
</s>
<s>
In	O
the	O
case	O
of	O
distributed	B-Operating_System
algorithms	I-Operating_System
,	O
computational	O
problems	O
are	O
typically	O
related	O
to	O
graphs	O
.	O
</s>
<s>
Often	O
the	O
graph	O
that	O
describes	O
the	O
structure	O
of	O
the	O
computer	B-Architecture
network	I-Architecture
is	O
the	O
problem	O
instance	O
.	O
</s>
<s>
The	O
computer	B-Application
program	I-Application
finds	O
a	O
coloring	O
of	O
the	O
graph	O
,	O
encodes	O
the	O
coloring	O
as	O
a	O
string	O
,	O
and	O
outputs	O
the	O
result	O
.	O
</s>
<s>
The	O
graph	O
G	O
is	O
the	O
structure	O
of	O
the	O
computer	B-Architecture
network	I-Architecture
.	O
</s>
<s>
There	O
is	O
one	O
computer	O
for	O
each	O
node	O
of	O
G	O
and	O
one	O
communication	O
link	O
for	O
each	O
edge	O
of	O
G	O
.	O
Initially	O
,	O
each	O
computer	O
only	O
knows	O
about	O
its	O
immediate	O
neighbors	O
in	O
the	O
graph	O
G	O
;	O
the	O
computers	O
must	O
exchange	O
messages	O
with	O
each	O
other	O
to	O
discover	O
more	O
about	O
the	O
structure	O
of	O
G	O
.	O
Each	O
computer	O
must	O
produce	O
its	O
own	O
color	O
as	O
output	B-General_Concept
.	O
</s>
<s>
The	O
main	O
focus	O
is	O
on	O
coordinating	O
the	O
operation	O
of	O
an	O
arbitrary	O
distributed	B-Architecture
system	I-Architecture
.	O
</s>
<s>
While	O
the	O
field	O
of	O
parallel	O
algorithms	O
has	O
a	O
different	O
focus	O
than	O
the	O
field	O
of	O
distributed	B-Operating_System
algorithms	I-Operating_System
,	O
there	O
is	O
much	O
interaction	O
between	O
the	O
two	O
fields	O
.	O
</s>
<s>
For	O
example	O
,	O
the	O
Cole	B-Operating_System
–	I-Operating_System
Vishkin	I-Operating_System
algorithm	I-Operating_System
for	O
graph	O
coloring	O
was	O
originally	O
presented	O
as	O
a	O
parallel	O
algorithm	O
,	O
but	O
the	O
same	O
technique	O
can	O
also	O
be	O
used	O
directly	O
as	O
a	O
distributed	B-Operating_System
algorithm	I-Operating_System
.	O
</s>
<s>
Moreover	O
,	O
a	O
parallel	O
algorithm	O
can	O
be	O
implemented	O
either	O
in	O
a	O
parallel	O
system	O
(	O
using	O
shared	B-Operating_System
memory	I-Operating_System
)	O
or	O
in	O
a	O
distributed	B-Architecture
system	I-Architecture
(	O
using	O
message	B-Architecture
passing	I-Architecture
)	O
.	O
</s>
<s>
The	O
traditional	O
boundary	O
between	O
parallel	O
and	O
distributed	B-Operating_System
algorithms	I-Operating_System
(	O
choose	O
a	O
suitable	O
network	O
vs.	O
run	O
in	O
any	O
given	O
network	O
)	O
does	O
not	O
lie	O
in	O
the	O
same	O
place	O
as	O
the	O
boundary	O
between	O
parallel	O
and	O
distributed	B-Architecture
systems	I-Architecture
(	O
shared	B-Operating_System
memory	I-Operating_System
vs.	O
message	B-Architecture
passing	I-Architecture
)	O
.	O
</s>
<s>
Indeed	O
,	O
often	O
there	O
is	O
a	O
trade-off	O
between	O
the	O
running	O
time	O
and	O
the	O
number	O
of	O
computers	O
:	O
the	O
problem	O
can	O
be	O
solved	O
faster	O
if	O
there	O
are	O
more	O
computers	O
running	O
in	O
parallel	O
(	O
see	O
speedup	B-Operating_System
)	O
.	O
</s>
<s>
In	O
the	O
analysis	O
of	O
distributed	B-Operating_System
algorithms	I-Operating_System
,	O
more	O
attention	O
is	O
usually	O
paid	O
on	O
communication	O
operations	O
than	O
computational	O
steps	O
.	O
</s>
<s>
Perhaps	O
the	O
simplest	O
model	O
of	O
distributed	B-Architecture
computing	I-Architecture
is	O
a	O
synchronous	O
system	O
where	O
all	O
nodes	B-Protocol
operate	O
in	O
a	O
lockstep	O
fashion	O
.	O
</s>
<s>
During	O
each	O
communication	O
round	O
,	O
all	O
nodes	B-Protocol
in	O
parallel	O
(	O
1	O
)	O
receive	O
the	O
latest	O
messages	O
from	O
their	O
neighbours	O
,	O
(	O
2	O
)	O
perform	O
arbitrary	O
local	O
computation	O
,	O
and	O
(	O
3	O
)	O
send	O
new	O
messages	O
to	O
their	O
neighbors	O
.	O
</s>
<s>
On	O
the	O
one	O
hand	O
,	O
any	O
computable	O
problem	O
can	O
be	O
solved	O
trivially	O
in	O
a	O
synchronous	O
distributed	B-Architecture
system	I-Architecture
in	O
approximately	O
2D	O
communication	O
rounds	O
:	O
simply	O
gather	O
all	O
information	O
in	O
one	O
location	O
(	O
D	O
rounds	O
)	O
,	O
solve	O
the	O
problem	O
,	O
and	O
inform	O
each	O
node	O
about	O
the	O
solution	O
(	O
D	O
rounds	O
)	O
.	O
</s>
<s>
On	O
the	O
other	O
hand	O
,	O
if	O
the	O
running	O
time	O
of	O
the	O
algorithm	O
is	O
much	O
smaller	O
than	O
D	O
communication	O
rounds	O
,	O
then	O
the	O
nodes	B-Protocol
in	O
the	O
network	O
must	O
produce	O
their	O
output	B-General_Concept
without	O
having	O
the	O
possibility	O
to	O
obtain	O
information	O
about	O
distant	O
parts	O
of	O
the	O
network	O
.	O
</s>
<s>
In	O
other	O
words	O
,	O
the	O
nodes	B-Protocol
must	O
make	O
globally	O
consistent	O
decisions	O
based	O
on	O
information	O
that	O
is	O
available	O
in	O
their	O
local	O
D-neighbourhood	O
.	O
</s>
<s>
Many	O
distributed	B-Operating_System
algorithms	I-Operating_System
are	O
known	O
with	O
the	O
running	O
time	O
much	O
smaller	O
than	O
D	O
rounds	O
,	O
and	O
understanding	O
which	O
problems	O
can	O
be	O
solved	O
by	O
such	O
algorithms	O
is	O
one	O
of	O
the	O
central	O
research	O
questions	O
of	O
the	O
field	O
.	O
</s>
<s>
Traditional	O
computational	O
problems	O
take	O
the	O
perspective	O
that	O
the	O
user	O
asks	O
a	O
question	O
,	O
a	O
computer	O
(	O
or	O
a	O
distributed	B-Architecture
system	I-Architecture
)	O
processes	B-Operating_System
the	O
question	O
,	O
then	O
produces	O
an	O
answer	O
and	O
stops	O
.	O
</s>
<s>
However	O
,	O
there	O
are	O
also	O
problems	O
where	O
the	O
system	O
is	O
required	O
not	O
to	O
stop	O
,	O
including	O
the	O
dining	O
philosophers	O
problem	O
and	O
other	O
similar	O
mutual	B-Operating_System
exclusion	I-Operating_System
problems	O
.	O
</s>
<s>
In	O
these	O
problems	O
,	O
the	O
distributed	B-Architecture
system	I-Architecture
is	O
supposed	O
to	O
continuously	O
coordinate	O
the	O
use	O
of	O
shared	O
resources	O
so	O
that	O
no	O
conflicts	O
or	O
deadlocks	B-Operating_System
occur	O
.	O
</s>
<s>
There	O
are	O
also	O
fundamental	O
challenges	O
that	O
are	O
unique	O
to	O
distributed	B-Architecture
computing	I-Architecture
,	O
for	O
example	O
those	O
related	O
to	O
fault-tolerance	B-General_Concept
.	O
</s>
<s>
Examples	O
of	O
related	O
problems	O
include	O
consensus	B-Operating_System
problems	I-Operating_System
,	O
Byzantine	B-Operating_System
fault	I-Operating_System
tolerance	I-Operating_System
,	O
and	O
self-stabilisation	B-Operating_System
.	O
</s>
<s>
Much	O
research	O
is	O
also	O
focused	O
on	O
understanding	O
the	O
asynchronous	O
nature	O
of	O
distributed	B-Architecture
systems	I-Architecture
:	O
</s>
<s>
Synchronizers	B-Operating_System
can	O
be	O
used	O
to	O
run	O
synchronous	O
algorithms	O
in	O
asynchronous	O
systems	O
.	O
</s>
<s>
Logical	B-Operating_System
clocks	I-Operating_System
provide	O
a	O
causal	O
happened-before	B-Operating_System
ordering	O
of	O
events	O
.	O
</s>
<s>
Coordinator	O
election	B-Operating_System
(	O
or	O
leader	B-Operating_System
election	I-Operating_System
)	O
is	O
the	O
process	B-Operating_System
of	O
designating	O
a	O
single	O
process	B-Operating_System
as	O
the	O
organizer	O
of	O
some	O
task	O
distributed	O
among	O
several	O
computers	O
(	O
nodes	B-Protocol
)	O
.	O
</s>
<s>
Before	O
the	O
task	O
is	O
begun	O
,	O
all	O
network	B-Protocol
nodes	I-Protocol
are	O
either	O
unaware	O
which	O
node	O
will	O
serve	O
as	O
the	O
"	O
coordinator	O
"	O
(	O
or	O
leader	O
)	O
of	O
the	O
task	O
,	O
or	O
unable	O
to	O
communicate	O
with	O
the	O
current	O
coordinator	O
.	O
</s>
<s>
After	O
a	O
coordinator	O
election	B-Operating_System
algorithm	O
has	O
been	O
run	O
,	O
however	O
,	O
each	O
node	O
throughout	O
the	O
network	O
recognizes	O
a	O
particular	O
,	O
unique	O
node	O
as	O
the	O
task	O
coordinator	O
.	O
</s>
<s>
The	O
network	B-Protocol
nodes	I-Protocol
communicate	O
among	O
themselves	O
in	O
order	O
to	O
decide	O
which	O
of	O
them	O
will	O
get	O
into	O
the	O
"	O
coordinator	O
"	O
state	O
.	O
</s>
<s>
For	O
example	O
,	O
if	O
each	O
node	O
has	O
unique	O
and	O
comparable	O
identities	O
,	O
then	O
the	O
nodes	B-Protocol
can	O
compare	O
their	O
identities	O
,	O
and	O
decide	O
that	O
the	O
node	O
with	O
the	O
highest	O
identity	O
is	O
the	O
coordinator	O
.	O
</s>
<s>
The	O
definition	O
of	O
this	O
problem	O
is	O
often	O
attributed	O
to	O
LeLann	O
,	O
who	O
formalized	O
it	O
as	O
a	O
method	O
to	O
create	O
a	O
new	O
token	O
in	O
a	O
token	O
ring	B-Architecture
network	I-Architecture
in	O
which	O
the	O
token	O
has	O
been	O
lost	O
.	O
</s>
<s>
Coordinator	O
election	B-Operating_System
algorithms	O
are	O
designed	O
to	O
be	O
economical	O
in	O
terms	O
of	O
total	O
bytes	B-Application
transmitted	O
,	O
and	O
time	O
.	O
</s>
<s>
The	O
algorithm	O
suggested	O
by	O
Gallager	O
,	O
Humblet	O
,	O
and	O
Spira	O
for	O
general	O
undirected	O
graphs	O
has	O
had	O
a	O
strong	O
impact	O
on	O
the	O
design	O
of	O
distributed	B-Operating_System
algorithms	I-Operating_System
in	O
general	O
,	O
and	O
won	O
the	O
Dijkstra	O
Prize	O
for	O
an	O
influential	O
paper	O
in	O
distributed	B-Architecture
computing	I-Architecture
.	O
</s>
<s>
A	O
general	O
method	O
that	O
decouples	O
the	O
issue	O
of	O
the	O
graph	O
family	O
from	O
the	O
design	O
of	O
the	O
coordinator	O
election	B-Operating_System
algorithm	O
was	O
suggested	O
by	O
Korach	O
,	O
Kutten	O
,	O
and	O
Moran	O
.	O
</s>
<s>
In	O
order	O
to	O
perform	O
coordination	O
,	O
distributed	B-Architecture
systems	I-Architecture
employ	O
the	O
concept	O
of	O
coordinators	O
.	O
</s>
<s>
The	O
coordinator	O
election	B-Operating_System
problem	O
is	O
to	O
choose	O
a	O
process	B-Operating_System
from	O
among	O
a	O
group	O
of	O
processes	B-Operating_System
on	O
different	O
processors	O
in	O
a	O
distributed	B-Architecture
system	I-Architecture
to	O
act	O
as	O
the	O
central	O
coordinator	O
.	O
</s>
<s>
Several	O
central	O
coordinator	O
election	B-Operating_System
algorithms	O
exist	O
.	O
</s>
<s>
So	O
far	O
the	O
focus	O
has	O
been	O
on	O
designing	O
a	O
distributed	B-Architecture
system	I-Architecture
that	O
solves	O
a	O
given	O
problem	O
.	O
</s>
<s>
A	O
complementary	O
research	O
problem	O
is	O
studying	O
the	O
properties	O
of	O
a	O
given	O
distributed	B-Architecture
system	I-Architecture
.	O
</s>
<s>
The	O
halting	O
problem	O
is	O
an	O
analogous	O
example	O
from	O
the	O
field	O
of	O
centralised	O
computation	O
:	O
we	O
are	O
given	O
a	O
computer	B-Application
program	I-Application
and	O
the	O
task	O
is	O
to	O
decide	O
whether	O
it	O
halts	O
or	O
runs	O
forever	O
.	O
</s>
<s>
The	O
halting	O
problem	O
is	O
undecidable	O
in	O
the	O
general	O
case	O
,	O
and	O
naturally	O
understanding	O
the	O
behaviour	O
of	O
a	O
computer	B-Architecture
network	I-Architecture
is	O
at	O
least	O
as	O
hard	O
as	O
understanding	O
the	O
behaviour	O
of	O
one	O
computer	O
.	O
</s>
<s>
In	O
particular	O
,	O
it	O
is	O
possible	O
to	O
reason	O
about	O
the	O
behaviour	O
of	O
a	O
network	O
of	O
finite-state	B-Architecture
machines	I-Architecture
.	O
</s>
<s>
One	O
example	O
is	O
telling	O
whether	O
a	O
given	O
network	O
of	O
interacting	O
(	O
asynchronous	O
and	O
non-deterministic	O
)	O
finite-state	B-Architecture
machines	I-Architecture
can	O
reach	O
a	O
deadlock	B-Operating_System
.	O
</s>
