<s>
Druid	B-Application
is	O
a	O
column-oriented	B-General_Concept
,	O
open-source	B-Application
,	O
distributed	B-Operating_System
data	B-General_Concept
store	I-General_Concept
written	O
in	O
Java	B-Language
.	O
</s>
<s>
Druid	B-Application
is	O
designed	O
to	O
quickly	O
ingest	O
massive	O
quantities	O
of	O
event	O
data	O
,	O
and	O
provide	O
low-latency	O
queries	O
on	O
top	O
of	O
the	O
data	O
.	O
</s>
<s>
The	O
name	O
Druid	B-Application
comes	O
from	O
the	O
shapeshifting	O
Druid	B-Application
class	O
in	O
many	O
role-playing	O
games	O
,	O
to	O
reflect	O
that	O
the	O
architecture	O
of	O
the	O
system	O
can	O
shift	O
to	O
solve	O
different	O
types	O
of	O
data	O
problems	O
.	O
</s>
<s>
Druid	B-Application
is	O
commonly	O
used	O
in	O
business	O
intelligence-OLAP	O
applications	O
to	O
analyze	O
high	O
volumes	O
of	O
real-time	B-General_Concept
and	O
historical	O
data	O
.	O
</s>
<s>
Druid	B-Application
is	O
used	O
in	O
production	O
by	O
technology	O
companies	O
such	O
as	O
Alibaba	O
,	O
Airbnb	O
,	O
Cisco	O
,	O
eBay	B-Application
,	O
Lyft	B-Application
,	O
Netflix	O
,	O
PayPal	O
,	O
Pinterest	B-Application
,	O
Reddit	B-Application
,	O
Twitter	B-Application
,	O
Walmart	O
,	O
Wikimedia	O
Foundation	O
and	O
Yahoo	B-Application
.	O
</s>
<s>
Druid	B-Application
was	O
started	O
in	O
2011	O
by	O
Eric	O
Tschetter	O
,	O
Fangjin	O
Yang	O
,	O
Gian	B-Application
Merlino	I-Application
and	O
Vadim	O
Ogievetsky	O
to	O
power	O
the	O
analytics	O
product	O
of	O
Metamarkets	O
.	O
</s>
<s>
Fully	O
deployed	O
,	O
Druid	B-Application
runs	O
as	O
a	O
cluster	O
of	O
specialized	O
processes	O
(	O
called	O
nodes	O
in	O
Druid	B-Application
)	O
to	O
support	O
a	O
fault-tolerant	B-General_Concept
architecture	O
where	O
data	O
is	O
stored	O
redundantly	O
,	O
and	O
there	O
is	O
no	O
single	O
point	O
of	O
failure	O
.	O
</s>
<s>
The	O
cluster	O
includes	O
external	O
dependencies	O
for	O
coordination	O
(	O
Apache	B-Language
ZooKeeper	I-Language
)	O
,	O
metadata	O
storage	O
(	O
e.g.	O
</s>
<s>
MySQL	B-Application
,	O
PostgreSQL	B-Application
,	O
or	O
Derby	B-Language
)	O
,	O
and	O
a	O
deep	O
storage	O
facility	O
(	O
e.g.	O
</s>
<s>
HDFS	O
,	O
or	O
Amazon	B-General_Concept
S3	I-General_Concept
)	O
for	O
permanent	O
data	O
backup	O
.	O
</s>
<s>
Client	O
queries	O
first	O
hit	O
broker	O
nodes	O
,	O
which	O
forward	O
them	O
to	O
the	O
appropriate	O
data	O
nodes	O
(	O
either	O
historical	O
or	O
real-time	B-General_Concept
)	O
.	O
</s>
<s>
Since	O
Druid	B-Application
segments	O
may	O
be	O
partitioned	O
,	O
an	O
incoming	O
query	O
can	O
require	O
data	O
from	O
multiple	O
segments	O
and	O
partitions	O
(	O
or	O
shards	B-General_Concept
)	O
stored	O
on	O
different	O
nodes	O
in	O
the	O
cluster	O
.	O
</s>
<s>
Apache	B-Language
ZooKeeper	I-Language
is	O
used	O
to	O
register	O
all	O
nodes	O
,	O
manage	O
certain	O
aspects	O
of	O
internode	O
communications	O
,	O
and	O
provide	O
for	O
leader	O
elections	O
.	O
</s>
<s>
Researchers	O
have	O
compared	O
the	O
performance	O
of	O
Hive	B-General_Concept
,	O
Presto	B-Language
,	O
and	O
Druid	B-Application
using	O
a	O
denormalized	O
Star	B-Application
Schema	I-Application
Benchmark	O
based	O
on	O
the	O
TPC-H	O
standard	O
.	O
</s>
<s>
Druid	B-Application
was	O
tested	O
using	O
both	O
a	O
“	O
Druid	B-Application
Best	O
”	O
configuration	O
using	O
tables	O
with	O
hashed	O
partitions	O
and	O
a	O
“	O
Druid	B-Application
Suboptimal	O
”	O
configuration	O
which	O
does	O
not	O
use	O
hashed	O
partitions	O
.	O
</s>
<s>
Druid	B-Application
performance	O
was	O
measured	O
as	O
at	O
least	O
98%	O
faster	O
than	O
Hive	B-General_Concept
and	O
at	O
least	O
90%	O
faster	O
than	O
Presto	B-Language
in	O
each	O
scenario	O
,	O
even	O
when	O
using	O
the	O
Druid	B-Application
Suboptimized	O
configuration	O
.	O
</s>
