Introduction to Apache Spark

An Introduction to
Apache Spark
Anastasios Skarlatidis
@anskarl
Software Engineer/Researcher
IIT, NCSR "Demokritos"

• Part I: Getting to know Spark
• Part II: Basic programming
• Part III: Spark under the hood
• Part IV: Advanced features
Outline

Spark in a Nutshell
• General cluster computing platform:
• Distributed in-memory computational framework.
• SQL, Machine Learning, Stream Processing, etc.
• Easy to use, powerful, high-level API:
• Scala, Java, Python and R.

Unified Stack
Spark SQL
Spark
Streaming
(real-time
processing)
MLlib
(Machine
Learning)
GraphX
(graph
processing)
Spark Core
Standalone Scheduler YARN Mesos

High Performance
• In-memory cluster computing.
• Ideal for iterative algorithms.
• Faster than Hadoop:
• 10x on disk.
• 100x in memory.

Brief History
• Originally developed in 2009, UC Berkeley AMP Lab.
• Open-sourced in 2010.
• As of 2014, Spark is a top-level Apache project.
• Fastest open-source engine for sorting 100 ΤΒ:
• Won the 2014 Daytona GraySort contest.
• Throughput: 4.27 TB/min

Who uses Spark,
and for what?
A. Data Scientists:
• Analyze and model data.
• Data transformations and prototyping.
• Statistics and Machine Learning.
B. Software Engineers:
• Implement production data processing systems.
• Require a reasonable API for distributed processing.
• Reliable, high performance, easy to monitor platform.

Resilient Distributed Dataset
RDD is an immutable and partitioned collection:
• Resilient: it can be recreated, when data in memory is lost.
• Distributed: stored in memory across the cluster.
• Dataset: data that comes from file or created
programmatically.
RDD
partitions

Resilient Distributed Datasets
• Feels like coding using typical Scala collections.
• RDD can be build:
1. Directly from a datasource (e.g., text file, HDFS, etc.),
2. or by applying a transformation to another RDD(s).
• Main features:
• RDDs are computed lazily.
• Automatically rebuild on failure.
• Persistence for reuse (RAM and/or disk).

Spark Shell
$
cd
spark

$
./bin/spark-‐shell

Spark
assembly
has
been
built
with
Hive,
including
Datanucleus
jars
on
classpath

Welcome
to

____

__

/
__/__

___
_____/
/__

_
/
_
/
_
`/
__/

'_/

/___/
.__/_,_/_/
/_/_

version
1.2.1

/_/

Using
Scala
version
2.10.4
(Java
HotSpot(TM)
64-‐Bit
Server
VM,
Java
1.7.0_71)

Type
in
expressions
to
have
them
evaluated.

Type
:help
for
more
information.

Spark
context
available
as
sc.

scala>

Standalone Applications
Sbt:

"org.apache.spark"
%%
"spark-‐core"
%
"1.2.1"
Maven:

groupId:
org.apache.spark

artifactId:
spark-‐core_2.10

version:
1.2.1

Initiate Spark Context
import
org.apache.spark.SparkContext

import
org.apache.spark.SparkContext._

import
org.apache.spark.SparkConf

object
SimpleApp
extends
App
{

val
conf
=
new
SparkConf().setAppName("Hello
Spark")

val
sc
=
new
SparkContext(conf)

}

Rich, High-level API
map

filter

sort

groupBy

union

join

…

reduce

count

fold

reduceByKey

groupByKey

cogroup

zip

…

sample

take

first

partitionBy

mapWith

pipe

save

…

Loading and Saving
• File Systems: Local FS, Amazon S3 and HDFS.
• Supported formats: Text files, JSON, Hadoop sequence files,
parquet files, protocol buffers and object files.
• Structured data with Spark SQL: Hive, JSON, JDBC,
Cassandra, HBase and ElasticSearch.

Create RDDs
//
sc:
SparkContext
instance

//
Scala
List
to
RDD

val
rdd0
=
sc.parallelize(List(1,
2,
3,
4))

//
Load
lines
of
a
text
file

val
rdd1
=
sc.textFile("path/to/filename.txt")

//
Load
a
file
from
HDFS

val
rdd2
=
sc.hadoopFile("hdfs://master:port/path")

//
Load
lines
of
a
compressed
text
file

val
rdd3
=
sc.textFile("file:///path/to/compressedText.gz")

//
Load
lines
of
multiple
files

val
rdd4
=
sc.textFile("s3n://log-‐files/2014/*.log")

RDD Operations
1. Transformations: define new RDDs based on current one,
e.g., filter, map, reduce, groupBy, etc.
RDD New RDD
2. Actions: return values, e.g., count, sum, collect, etc.
value
RDD

Transformations (I): basics
val
nums
=
2,
3))

//
Pass
each
element
through
a
function

val
squares
=
nums.map(x
=>
x
*
x)
//{1,
4,
9}

//
Keep
elements
passing
a
predicate

val
even
=
squares.filter(_
%
2
==
0)
//{4}

//
Map
each
element
to
zero
or
more
others

val
mn
=
nums.flatMap(x
=>
1
to
x)
//{1,
1,
2,
1,
2,
3}

Transformations (I): illustrated
nums
squares
ParallelCollectionRDD
even
mn
nums.flatMap(x
=>
1
to
x)
squares.filter(_
%
2
==
0)
MappedRDD
FilteredRDD
nums.map(x
=>
x
*
x)
FlatMappedRDD

Transformations (II): key - value
val
pets
=
sc.parallelize(List(("cat",
1),
("dog",
1),

("cat",
2)))

ValueKey
pets.filter{case
(k,
v)
=>
k
==
"cat"}

//
{(cat,1),
(cat,2)}

pets.map{case
(k,
v)
=>
(k,
v
+
1)}

//
{(cat,2),
(dog,2),
(cat,3)}

pets.mapValues(v
=>
v
+
1)

//
{(cat,2),
(dog,2),
(cat,3)}

Transformations (II): key - value
//
Aggregation

pets.reduceByKey((l,
r)
=>
l
+
r)
//{(cat,3),
(dog,1)}

//
Grouping

pets.groupByKey()
//{(cat,
Seq(1,
2)),
(dog,
Seq(1)}

//
Sorting

pets.sortByKey()

//{(cat,
1),
(cat,
2),
(dog,
1)}

val
pets
=
sc.parallelize(List(("cat",
1),
("dog",
1),

("cat",
2)))

ValueKey

Transformations (III): key - value
//RDD[(URL,
page_name)]
tuples

val
names
=
sc.textFile("names.txt").map(…)…

//RDD[(URL,
visit_counts)]
tuples

val
visits
=
sc.textFile("counts.txt").map(…)…

//RDD[(URL,
(visit
counts,
page
name))]

val
joined
=
visits.join(names)

Basics: Actions
val
nums
=
2,
3))

//
Count
number
of
elements

nums.count()

//
=
3

//
Merge
with
an
associative
function

nums.reduce((l,
r)
=>
l
+
r)

//
=
6

//
Write
elements
to
a
text
file

nums.saveAsTextFile("path/to/filename.txt")

Workflow
data transformation action
result

Part III:
Spark under the hood

1. Job: work required to compute an RDD.
2. Each job is divided to stages.
3. Task:
• Unit of work within a stage
• Corresponds to one RDD partition.
Units of Execution Model
Task 0
Job
Stage 0
…Task 1 Task 0
Stage 1
…Task 1 …

Execution Model
Driver Program
SparkContext
Worker Node
Task Task
Executor
Worker Node
Task Task
Executorval
lines
=
sc.textFile("README.md")

val
countedLines
=
lines.count()

Example: word count
val
lines
=
sc.textFile("hamlet.txt")

val
counts
=
lines.flatMap(_.split("
"))

//
(a)

.map(word
=>
(word,
1))

//
(b)

.reduceByKey(_
+
_)

//
(c)
"to
be
or"
"not
to
be"
(a) "to"

"be"

"or"
"not"

"to"

"be"
("to",
1)

("be",
1)

("or",
1)
("not",
1)

("to",
1)

("be",
1)
(b)
("be",
2)

("not",
1)
("or",
1)

("to",
2)
(c)
to be or
not to be

12:
val
lines
=
//
HadoopRDD[0],
MappedRDD[1]

13:

14:
val
counts
=
"))
//
FlatMappedRDD[2]

15:

.map(word
=>
(word,
1))

//
MappedRDD[3]

16:

.reduceByKey(_
+
_)

//
ShuffledRDD[4]

17:

18:
counts.toDebugString

res0:
String
=

(2)
ShuffledRDD[4]
at
reduceByKey
at
<console>:16
[]

+-‐(2)
MappedRDD[3]
at
map
at
<console>:15
[]

|

FlatMappedRDD[2]
at
flatMap
at
<console>:14
[]

|

hamlet.txt
MappedRDD[1]
at
textFile
at
<console>:12
[]

|

hamlet.txt
HadoopRDD[0]
at
textFile
at
<console>:12
[]

Visualize an RDD

Lineage Graph
val
lines
=
//
MappedRDD[1],
HadoopRDD[0]

val
counts
=
"))

//
FlatMappedRDD[2]

.map(word
=>
(word,
1))
//
MappedRDD[3]

.reduceByKey(_
+
_)

//
ShuffledRDD[4]
[0] [1] [2] [3] [4]
HadoopRDD MappedRDD FlatMappedRDD MappedRDD ShuffledRDD

Lineage Graph
val
lines
=
//
MappedRDD[1],
HadoopRDD[0]

val
counts
=
"))

//
FlatMappedRDD[2]

.map(word
=>
(word,
1))
//
MappedRDD[3]

.reduceByKey(_
+
_)

//
ShuffledRDD[4]

Execution Plan
val
lines
=
//
MappedRDD[1],
HadoopRDD[0]

val
counts
=
"))

//
FlatMappedRDD[2]

.map(word
=>
(word,
1))
//
MappedRDD[3]

.reduceByKey(_
+
_)

//
ShuffledRDD[4]
Stage 1 Stage 2
pipelining

Persistence
• When we use the same RDD multiple times:
• Spark will recompute the RDD.
• Expensive to iterative algorithms.
• Spark can persist RDDs, avoiding recomputations.

Levels of persistence
val
result
=
input.map(expensiveComputation)

result.persist(LEVEL)
LEVEL
Space
Consumption
CPU time In memory On disk
MEMORY_ONLY (default) High Low Y N
MEMORY_ONLY_SER Low High Y N
MEMORY_AND_DISK High Medium Some Some
MEMORY_AND_DISK_SER Low High Some Some
DISK_ONLY Low High N Y

Persistence Behaviour
• Each node will store its computed partition.
• In case of a failure, Spark recomputes the
missing partitions.
• Least Recently Used cache policy:
• Memory-only: recompute partitions.
• Memory-and-disk: recompute and write to disk.
• Manually remove from cache: unpersist()

Shared Variables
1. Accumulators: aggregate values from worker
nodes back to the driver program.
2. Broadcast variables: distribute values to all
worker nodes.

Accumulator Example
val
input
=
sc.textFile("input.txt")

val
sum
=
sc.accumulator(0)

val
count
=
sc.accumulator(0)

input

.filter(line
=>
line.size
>
0)

.flatMap(line
=>
line.split("
"))

.map(word
=>
word.size)

.foreach{

size
=>

sum
+=
size
//
increment
accumulator

count
+=
1

//
increment
accumulator

}

val
average
=
sum.value.toDouble
/
count.value
driver only
initialize the
accumulators

• Safe: Updates inside actions will only applied once.
• Unsafe: Updates inside transformation may applied
more than once!!!
Accumulators and Fault
Tolerance

Broadcast Variables
• Closures and the variables they use are send
separately to each task.
• We may want to share some variable (e.g., a Map)
across tasks/operations.
• This can efficiently done with broadcast variables.

Example without
broadcast variables
//
RDD[(String,
String)]

val
names
=
…
//load
(URL,
page
name)
tuples

//
RDD[(String,
Int)]

val
visits
=
…
//load
(URL,
visit
counts)
tuples

//
Map[String,
String]

val
pageMap
=
names.collect.toMap

val
joined
=
visits.map{

case
(url,
counts)
=>

(url,
(pageMap(url),
counts))

}
pageMap
is
sent
along

with
every
task

Example with
broadcast variables
//
RDD[(String,
String)]

val
names
=
…
//load
(URL,
page
name)
tuples

//
RDD[(String,
Int)]

val
visits
=
…
//load
(URL,
visit
counts)
tuples

//
Map[String,
String]

val
pageMap
=
names.collect.toMap

val
bcMap
=
sc.broadcast(pageMap)

val
joined
=
visits.map{

case
(url,
counts)
=>

(url,
(bcMap.value(url),
counts))

}
Broadcast variable
pageMap
is
sent
only

to
each
node
once

Staging
groupBy
map filter
join

Staging
groupBy
map filter
join
Caching

Staging
groupBy
map filter
join
Stage 1
Stage 2
Stage 3

Introduction to Apache Spark

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Introduction to Apache Spark (20)

Recently uploaded (20)

Introduction to Apache Spark