<s>
Performance	B-Device
tuning	I-Device
is	O
the	O
improvement	O
of	O
system	O
performance	O
.	O
</s>
<s>
Most	O
systems	O
will	O
respond	O
to	O
increased	O
load	B-Operating_System
with	O
some	O
degree	O
of	O
decreasing	O
performance	O
.	O
</s>
<s>
A	O
system	O
's	O
ability	O
to	O
accept	O
higher	O
load	B-Operating_System
is	O
called	O
scalability	B-Architecture
,	O
and	O
modifying	O
a	O
system	O
to	O
handle	O
a	O
higher	O
load	B-Operating_System
is	O
synonymous	O
to	O
performance	B-Device
tuning	I-Device
.	O
</s>
<s>
This	O
usually	O
occurs	O
because	O
high	O
system	O
loading	B-Operating_System
,	O
causing	O
some	O
part	O
of	O
the	O
system	O
to	O
reach	O
a	O
limit	O
in	O
its	O
ability	O
to	O
respond	O
.	O
</s>
<s>
Among	O
them	O
are	O
code	O
optimization	O
,	O
load	B-Application
balancing	I-Application
,	O
caching	B-General_Concept
strategy	O
,	O
distributed	B-Architecture
computing	I-Architecture
and	O
self-tuning	O
.	O
</s>
<s>
Performance	B-General_Concept
engineering	I-General_Concept
is	O
the	O
discipline	O
encompassing	O
roles	O
,	O
skills	O
,	O
activities	O
,	O
practices	O
,	O
tools	O
,	O
and	O
deliverables	O
used	O
to	O
meet	O
the	O
non-functional	O
requirements	O
of	O
a	O
designed	O
system	O
,	O
such	O
as	O
increase	O
business	O
revenue	O
,	O
reduction	O
of	O
system	O
failure	O
,	O
delayed	O
projects	O
,	O
and	O
avoidance	O
of	O
unnecessary	O
usage	O
of	O
resources	O
or	O
work	O
.	O
</s>
<s>
Elaboration	O
of	O
the	O
processes	O
in	O
use	B-Architecture
cases	I-Architecture
and	O
system	O
volumetrics	O
.	O
</s>
<s>
System	O
construction	O
,	O
including	O
performance	B-Device
tuning	I-Device
.	O
</s>
<s>
Some	O
optimizations	O
include	O
improving	O
the	O
code	O
so	O
that	O
work	O
is	O
done	O
once	O
before	O
a	O
loop	O
rather	O
than	O
inside	O
a	O
loop	O
or	O
replacing	O
a	O
call	O
to	O
a	O
simple	O
selection	B-Algorithm
sort	I-Algorithm
with	O
a	O
call	O
to	O
the	O
more	O
complicated	O
algorithm	O
for	O
a	O
quicksort	B-Algorithm
.	O
</s>
<s>
Caching	B-General_Concept
is	O
a	O
fundamental	O
method	O
of	O
removing	O
performance	O
bottlenecks	O
that	O
are	O
the	O
result	O
of	O
slow	O
access	O
to	O
data	O
.	O
</s>
<s>
Caching	B-General_Concept
improves	O
performance	O
by	O
retaining	O
frequently	O
used	O
information	O
in	O
high	O
speed	O
memory	O
,	O
reducing	O
access	O
time	O
and	O
avoiding	O
repeated	O
computation	O
.	O
</s>
<s>
Caching	B-General_Concept
is	O
an	O
effective	O
manner	O
of	O
improving	O
performance	O
in	O
situations	O
where	O
the	O
principle	O
of	O
locality	B-General_Concept
of	I-General_Concept
reference	I-General_Concept
applies	O
.	O
</s>
<s>
The	O
methods	O
used	O
to	O
determine	O
which	O
data	O
is	O
stored	O
in	O
progressively	O
faster	O
storage	O
are	O
collectively	O
called	O
caching	B-General_Concept
strategies	O
.	O
</s>
<s>
Examples	O
are	O
ASP.NET	B-Application
cache	I-Application
,	O
CPU	B-General_Concept
cache	I-General_Concept
,	O
etc	O
.	O
</s>
<s>
Arranging	O
so	O
all	O
systems	O
are	O
used	O
equally	O
is	O
referred	O
to	O
as	O
load	B-Application
balancing	I-Application
and	O
can	O
improve	O
overall	O
performance	O
.	O
</s>
<s>
Load	B-Application
balancing	I-Application
is	O
often	O
used	O
to	O
achieve	O
further	O
gains	O
from	O
a	O
distributed	B-Architecture
system	I-Architecture
by	O
intelligently	O
selecting	O
which	O
machine	O
to	O
run	O
an	O
operation	O
on	O
based	O
on	O
how	O
busy	O
all	O
potential	O
candidates	O
are	O
,	O
and	O
how	O
well	O
suited	O
each	O
machine	O
is	O
to	O
the	O
type	O
of	O
operation	O
that	O
needs	O
to	O
be	O
performed	O
.	O
</s>
<s>
Distributed	B-Architecture
computing	I-Architecture
is	O
used	O
for	O
increasing	O
the	O
potential	O
for	O
parallel	O
execution	O
on	O
modern	O
CPU	O
architectures	O
continues	O
,	O
the	O
use	O
of	O
distributed	B-Architecture
systems	I-Architecture
is	O
essential	O
to	O
achieve	O
performance	O
benefits	O
from	O
the	O
available	O
parallelism	B-Operating_System
.	O
</s>
<s>
High-performance	O
cluster	B-Architecture
computing	I-Architecture
is	O
a	O
well-known	O
use	O
of	O
distributed	B-Architecture
systems	I-Architecture
for	O
performance	O
improvements	O
.	O
</s>
<s>
Distributed	B-Architecture
computing	I-Architecture
and	O
clustering	O
can	O
negatively	O
impact	O
latency	O
while	O
simultaneously	O
increasing	O
load	B-Operating_System
on	O
shared	O
resources	O
,	O
such	O
as	O
database	O
systems	O
.	O
</s>
<s>
To	O
minimize	O
latency	O
and	O
avoid	O
bottlenecks	O
,	O
distributed	B-Architecture
computing	I-Architecture
can	O
benefit	O
significantly	O
from	O
distributed	O
caches	B-General_Concept
.	O
</s>
