<s>
In	O
computer	B-General_Concept
science	I-General_Concept
,	O
high-throughput	B-Operating_System
computing	I-Operating_System
(	O
HTC	O
)	O
is	O
the	O
use	O
of	O
many	O
computing	O
resources	O
over	O
long	O
periods	O
of	O
time	O
to	O
accomplish	O
a	O
computational	O
task	O
.	O
</s>
<s>
This	O
research	O
is	O
similar	O
to	O
transaction	B-General_Concept
processing	I-General_Concept
,	O
but	O
at	O
a	O
much	O
larger	O
and	O
distributed	O
scale	O
.	O
</s>
<s>
Some	O
HTC	O
systems	O
,	O
such	O
as	O
HTCondor	B-Operating_System
and	O
PBS	B-Application
,	O
can	O
run	O
tasks	O
on	O
opportunistic	O
resources	O
.	O
</s>
<s>
There	O
are	O
many	O
differences	O
between	O
high-throughput	B-Operating_System
computing	I-Operating_System
,	O
high-performance	B-Architecture
computing	I-Architecture
(	O
HPC	O
)	O
,	O
and	O
many-task	B-Operating_System
computing	I-Operating_System
(	O
MTC	O
)	O
.	O
</s>
<s>
As	O
an	O
alternative	O
definition	O
,	O
the	O
European	O
Grid	O
Infrastructure	O
defines	O
HTC	O
as	O
"	O
a	O
computing	O
paradigm	O
that	O
focuses	O
on	O
the	O
efficient	O
execution	O
of	O
a	O
large	O
number	O
of	O
loosely-coupled	O
tasks	O
"	O
,	O
while	O
HPC	O
systems	O
tend	O
to	O
focus	O
on	O
tightly	O
coupled	O
parallel	B-Operating_System
jobs	O
,	O
and	O
as	O
such	O
they	O
must	O
execute	O
within	O
a	O
particular	O
site	O
with	O
low-latency	O
interconnects	O
.	O
</s>
<s>
HTC	O
systems	O
achieve	O
this	O
using	O
various	O
grid	B-Architecture
computing	I-Architecture
technologies	O
and	O
techniques	O
.	O
</s>
