<s>
Warewulf	B-Operating_System
is	O
a	O
computer	B-Architecture
cluster	I-Architecture
implementation	O
toolkit	B-Library
that	O
facilitates	O
the	O
process	O
of	O
installing	B-Application
a	O
cluster	B-Architecture
and	O
long	O
term	O
administration	B-Application
.	O
</s>
<s>
It	O
does	O
this	O
by	O
changing	O
the	O
administration	B-Application
paradigm	O
to	O
make	O
all	O
of	O
the	O
slave	B-Operating_System
node	B-Protocol
file	B-Application
systems	I-Application
manageable	O
from	O
one	O
point	O
,	O
and	O
automate	O
the	O
distribution	O
of	O
the	O
node	B-Protocol
file	B-Application
system	I-Application
during	O
node	B-Protocol
boot	O
.	O
</s>
<s>
It	O
allows	O
a	O
central	O
administration	B-Application
model	O
for	O
all	O
slave	B-Operating_System
nodes	O
and	O
includes	O
the	O
tools	O
needed	O
to	O
build	O
configuration	O
files	O
,	O
monitor	O
,	O
and	O
control	O
the	O
nodes	O
.	O
</s>
<s>
It	O
is	O
totally	O
customizable	O
and	O
can	O
be	O
adapted	O
to	O
just	O
about	O
any	O
type	O
of	O
cluster	B-Architecture
.	O
</s>
<s>
From	O
the	O
software	O
administration	B-Application
perspective	O
it	O
does	O
not	O
make	O
much	O
difference	O
if	O
you	O
are	O
running	O
2	O
nodes	O
or	O
500	O
nodes	O
.	O
</s>
<s>
The	O
procedure	O
is	O
still	O
the	O
same	O
,	O
which	O
is	O
why	O
Warewulf	B-Operating_System
is	O
scalable	O
from	O
the	O
admins	O
perspective	O
.	O
</s>
<s>
Also	O
,	O
because	O
it	O
uses	O
a	O
standard	O
chroot'able	O
file	B-Application
system	I-Application
for	O
every	O
node	B-Protocol
,	O
it	O
is	O
extremely	O
configurable	O
and	O
lends	O
itself	O
to	O
custom	O
environments	O
very	O
easily	O
.	O
</s>
<s>
While	O
Warewulf	B-Operating_System
was	O
designed	O
to	O
be	O
a	O
high-performance	B-Architecture
computing	I-Architecture
(	O
HPC	O
)	O
system	O
,	O
it	O
is	O
not	O
an	O
HPC	O
system	O
in	O
itself	O
.	O
</s>
<s>
Warewulf	B-Operating_System
is	O
more	O
along	O
the	O
lines	O
of	O
a	O
distributed	O
Linux	B-Application
distribution	O
,	O
or	O
more	O
specifically	O
a	O
system	O
for	O
replicating	O
and	O
managing	O
small	O
,	O
lightweight	O
Linux	B-Application
systems	O
from	O
one	O
master	O
.	O
</s>
<s>
Using	O
Warewulf	B-Operating_System
,	O
HPC	O
packages	O
such	O
as	O
LAM/MPI/MPICH	O
,	O
Sun	B-Operating_System
Grid	I-Operating_System
Engine	I-Operating_System
,	O
PVM	B-Operating_System
,	O
etc	O
.	O
</s>
<s>
can	O
be	O
easily	O
deployed	O
throughout	O
the	O
cluster	B-Architecture
.	O
</s>
<s>
Warewulf	B-Operating_System
solves	O
the	O
problem	O
of	O
slave	B-Operating_System
node	B-Protocol
management	O
rather	O
than	O
being	O
a	O
strict	O
HPC	O
specific	O
system	O
(	O
even	O
though	O
it	O
was	O
designed	O
with	O
HPC	O
in	O
mind	O
)	O
.	O
</s>
<s>
Because	O
of	O
this	O
it	O
is	O
as	O
flexible	O
as	O
a	O
home	O
grown	O
cluster	B-Architecture
,	O
but	O
administratively	O
scales	O
very	O
well	O
.	O
</s>
<s>
As	O
a	O
result	O
of	O
this	O
flexibility	O
and	O
ease	O
of	O
customization	O
,	O
Warewulf	B-Operating_System
has	O
been	O
used	O
not	O
only	O
on	O
production	O
HPC	O
implementations	O
,	O
but	O
also	O
development	O
systems	O
like	O
KASY0	O
(	O
the	O
first	O
system	O
to	O
break	O
the	O
one	O
hundred	O
dollar	O
per	O
GFLOPS	O
barrier	O
)	O
,	O
and	O
non	O
HPC	O
systems	O
such	O
as	O
web	B-Application
server	I-Application
cluster	B-Architecture
farms	O
,	O
intrusion	O
detection	O
clusters	O
,	O
and	O
high-availability	B-General_Concept
clusters	I-General_Concept
.	O
</s>
