<s>
In	O
distributed	B-Architecture
computing	I-Architecture
,	O
a	O
single	B-Operating_System
system	I-Operating_System
image	I-Operating_System
(	O
SSI	O
)	O
cluster	B-Architecture
is	O
a	O
cluster	B-Architecture
of	O
machines	O
that	O
appears	O
to	O
be	O
one	O
single	O
system	O
.	O
</s>
<s>
The	O
concept	O
is	O
often	O
considered	O
synonymous	O
with	O
that	O
of	O
a	O
distributed	B-Operating_System
operating	I-Operating_System
system	I-Operating_System
,	O
but	O
a	O
single	O
image	O
may	O
be	O
presented	O
for	O
more	O
limited	O
purposes	O
,	O
just	O
job	B-General_Concept
scheduling	I-General_Concept
for	O
instance	O
,	O
which	O
may	O
be	O
achieved	O
by	O
means	O
of	O
an	O
additional	O
layer	O
of	O
software	O
over	O
conventional	O
operating	B-General_Concept
system	I-General_Concept
images	O
running	O
on	O
each	O
node	B-Protocol
.	O
</s>
<s>
Many	O
SSI	O
systems	O
provide	O
process	B-Operating_System
migration	I-Operating_System
.	O
</s>
<s>
Processes	O
may	O
start	O
on	O
one	O
node	B-Protocol
and	O
be	O
moved	O
to	O
another	O
node	B-Protocol
,	O
possibly	O
for	O
resource	B-Application
balancing	I-Application
or	O
administrative	O
reasons	O
.	O
</s>
<s>
As	O
processes	O
are	O
moved	O
from	O
one	O
node	B-Protocol
to	O
another	O
,	O
other	O
associated	O
resources	O
(	O
for	O
example	O
IPC	B-Operating_System
resources	O
)	O
may	O
be	O
moved	O
with	O
them	O
.	O
</s>
<s>
Some	O
SSI	O
systems	O
allow	O
checkpointing	B-General_Concept
of	O
running	O
processes	O
,	O
allowing	O
their	O
current	O
state	O
to	O
be	O
saved	O
and	O
reloaded	O
at	O
a	O
later	O
date	O
.	O
</s>
<s>
Checkpointing	B-General_Concept
can	O
be	O
seen	O
as	O
related	O
to	O
migration	O
,	O
as	O
migrating	O
a	O
process	O
from	O
one	O
node	B-Protocol
to	O
another	O
can	O
be	O
implemented	O
by	O
first	O
checkpointing	B-General_Concept
the	O
process	O
,	O
then	O
restarting	O
it	O
on	O
another	O
node	B-Protocol
.	O
</s>
<s>
Alternatively	O
checkpointing	B-General_Concept
can	O
be	O
considered	O
as	O
migration	O
to	O
disk	O
.	O
</s>
<s>
"	O
ps	O
"	O
,	O
"	O
kill	O
"	O
on	O
Unix	B-Application
like	O
systems	O
)	O
operate	O
on	O
all	O
processes	O
in	O
the	O
cluster	B-Architecture
.	O
</s>
<s>
This	O
may	O
be	O
achieved	O
by	O
a	O
simple	O
NFS	B-Protocol
server	O
,	O
shared	O
disk	O
devices	O
or	O
even	O
file	O
replication	O
.	O
</s>
<s>
The	O
advantage	O
of	O
a	O
single	O
root	O
view	O
is	O
that	O
processes	O
may	O
be	O
run	O
on	O
any	O
available	O
node	B-Protocol
and	O
access	O
needed	O
files	O
with	O
no	O
special	O
precautions	O
.	O
</s>
<s>
If	O
the	O
cluster	B-Architecture
implements	O
process	B-Operating_System
migration	I-Operating_System
a	O
single	O
root	O
view	O
enables	O
direct	O
accesses	O
to	O
the	O
files	O
from	O
the	O
node	B-Protocol
where	O
the	O
process	O
is	O
currently	O
running	O
.	O
</s>
<s>
Some	O
SSI	O
systems	O
provide	O
a	O
way	O
of	O
"	O
breaking	O
the	O
illusion	O
"	O
,	O
having	O
some	O
node-specific	O
files	O
even	O
in	O
a	O
single	O
root	O
.	O
</s>
<s>
HP	O
TruCluster	B-General_Concept
provides	O
a	O
"	O
context	O
dependent	O
symbolic	O
link	O
"	O
(	O
CDSL	O
)	O
which	O
points	O
to	O
different	O
files	O
depending	O
on	O
the	O
node	B-Protocol
that	O
accesses	O
it	O
.	O
</s>
<s>
HP	O
VMScluster	B-Operating_System
provides	O
a	O
search	O
list	O
logical	O
name	O
with	O
node	B-Protocol
specific	O
files	O
occluding	O
cluster	B-Architecture
shared	O
files	O
where	O
necessary	O
.	O
</s>
<s>
There	O
may	O
be	O
some	O
restrictions	O
on	O
the	O
kinds	O
of	O
accesses	O
allowed	O
(	O
For	O
example	O
,	O
OpenSSI	B-Operating_System
ca	O
n't	O
mount	O
disk	O
devices	O
from	O
one	O
node	B-Protocol
on	O
another	O
node	B-Protocol
)	O
.	O
</s>
<s>
Some	O
SSI	O
systems	O
allow	O
processes	O
on	O
different	O
nodes	O
to	O
communicate	O
using	O
inter-process	B-Operating_System
communications	I-Operating_System
mechanisms	O
as	O
if	O
they	O
were	O
running	O
on	O
the	O
same	O
machine	O
.	O
</s>
<s>
On	O
some	O
SSI	O
systems	O
this	O
can	O
even	O
include	O
shared	B-Operating_System
memory	I-Operating_System
(	O
can	O
be	O
emulated	O
with	O
Software	B-Operating_System
Distributed	I-Operating_System
shared	I-Operating_System
memory	I-Operating_System
)	O
.	O
</s>
<s>
In	O
most	O
cases	O
inter-node	O
IPC	B-Operating_System
will	O
be	O
slower	O
than	O
IPC	B-Operating_System
on	O
the	O
same	O
machine	O
,	O
possibly	O
drastically	O
slower	O
for	O
shared	B-Operating_System
memory	I-Operating_System
.	O
</s>
<s>
Some	O
SSI	O
systems	O
provide	O
a	O
"	O
cluster	B-Architecture
IP	O
address	O
"	O
,	O
a	O
single	O
address	O
visible	O
from	O
outside	O
the	O
cluster	B-Architecture
that	O
can	O
be	O
used	O
to	O
contact	O
the	O
cluster	B-Architecture
as	O
if	O
it	O
were	O
one	O
machine	O
.	O
</s>
<s>
This	O
can	O
be	O
used	O
for	O
load	B-Application
balancing	I-Application
inbound	O
calls	O
to	O
the	O
cluster	B-Architecture
,	O
directing	O
them	O
to	O
lightly	O
loaded	O
nodes	O
,	O
or	O
for	O
redundancy	O
,	O
moving	O
the	O
cluster	B-Architecture
address	O
from	O
one	O
machine	O
to	O
another	O
as	O
nodes	O
join	O
or	O
leave	O
the	O
cluster	B-Architecture
.	O
</s>
<s>
Examples	O
here	O
vary	O
from	O
commercial	O
platforms	O
with	O
scaling	O
capabilities	O
,	O
to	O
packages/frameworks	O
for	O
creating	O
distributed	B-Architecture
systems	I-Architecture
,	O
as	O
well	O
as	O
those	O
that	O
actually	O
implement	O
a	O
single	B-Operating_System
system	I-Operating_System
image	I-Operating_System
.	O
</s>
