1. How Apache Spark
fits into the
Big Data landscape
Stockholm Big Data
2014-11-13
meetup.com/The-Stockholm-Big-Data-Group/events/212782912/
Paco Nathan
@pacoid
1
3. What is Spark?
Developed in 2009 at UC Berkeley AMPLab, then
open sourced in 2010, Spark has since become
one of the largest OSS communities in big data,
with over 200 contributors in 50+ organizations
spark.apache.org
“Organizations that are looking at big data challenges –
including collection, ETL, storage, exploration and analytics –
should consider Spark for its in-memory performance and
the breadth of its model. It supports advanced analytics
solutions on Hadoop clusters, including the iterative model
required for machine learning and graph analysis.”
Gartner, Advanced Analytics and Data Science (2014)
3
5. What is Spark?
Spark Core is the general execution engine for the
Spark platform that other functionality is built atop:
!
• in-memory computing capabilities deliver speed
• general execution model supports wide variety
of use cases
• ease of development – native APIs in Java, Scala,
Python (+ SQL, Clojure, R)
5
6. What is Spark?
WordCount in 3 lines of Spark
WordCount in 50+ lines of Java MR
6
7. What is Spark?
Sustained exponential growth, as one of the most
active Apache projects ohloh.net/orgs/apache
7
8. TL;DR: Smashing The Previous Petabyte Sort Record
databricks.com/blog/2014/11/05/spark-officially-sets-
a-new-record-in-large-scale-sorting.html
8
10. A Brief History: Functional Programming for Big Data
Theory, Eight Decades Ago:
what can be computed?
Haskell Curry
haskell.org
Alonso Church
wikipedia.org
Praxis, Four Decades Ago:
algebra for applicative systems
John Backus
acm.org
David Turner
wikipedia.org
Reality, Two Decades Ago:
machine data from web apps
Pattie Maes
MIT Media Lab
10
11. A Brief History: Functional Programming for Big Data
circa late 1990s:
explosive growth e-commerce and machine data
implied that workloads could not fit on a single
computer anymore…
notable firms led the shift to horizontal scale-out
on clusters of commodity hardware, especially
for machine learning use cases at scale
11
13. Amazon
“Early Amazon: Splitting the website” – Greg Linden
glinden.blogspot.com/2006/02/early-amazon-splitting-
website.html
!
eBay
“The eBay Architecture” – Randy Shoup, Dan Pritchett
addsimplicity.com/adding_simplicity_an_engi/
2006/11/you_scaled_your.html
addsimplicity.com.nyud.net:8080/downloads/
eBaySDForum2006-11-29.pdf
!
Inktomi (YHOO Search)
“Inktomi’s Wild Ride” – Erik Brewer (0:05:31 ff)
youtu.be/E91oEn1bnXM
!
Google
“Underneath the Covers at Google” – Jeff Dean (0:06:54 ff)
youtu.be/qsan-GQaeyk
perspectives.mvdirona.com/2008/06/11/
JeffDeanOnGoogleInfrastructure.aspx
!
MIT Media Lab
“Social Information Filtering for Music Recommendation” – Pattie Maes
pubs.media.mit.edu/pubs/papers/32paper.ps
ted.com/speakers/pattie_maes.html
13
14. A Brief History: Functional Programming for Big Data
circa 2002:
mitigate risk of large distributed workloads lost
due to disk failures on commodity hardware…
Google File System
Sanjay Ghemawat, Howard Gobioff, Shun-Tak Leung
research.google.com/archive/gfs.html
!
MapReduce: Simplified Data Processing on Large Clusters
Jeffrey Dean, Sanjay Ghemawat
research.google.com/archive/mapreduce.html
14
15. A Brief History: Functional Programming for Big Data
2002
2004
MapReduce paper
2002
MapReduce @ Google
2004 2006 2008 2010 2012 2014
2006
Hadoop @ Yahoo!
2014
Apache Spark top-level
2010
Spark paper
2008
Hadoop Summit
15
18. A Brief History: Functional Programming for Big Data
MapReduce
Pregel Giraph
Dremel Drill
S4 Storm
F1
MillWheel
General Batch Processing Specialized Systems:
Impala
GraphLab
iterative, interactive, streaming, graph, etc.
Tez
MR doesn’t compose well for large applications,
and so specialized systems emerged as workarounds
18
19. A Brief History: Functional Programming for Big Data
circa 2010:
a unified engine for enterprise data workflows,
based on commodity hardware a decade later…
Spark: Cluster Computing with Working Sets
Matei Zaharia, Mosharaf Chowdhury,
Michael Franklin, Scott Shenker, Ion Stoica
people.csail.mit.edu/matei/papers/2010/hotcloud_spark.pdf
!
Resilient Distributed Datasets: A Fault-Tolerant Abstraction for
In-Memory Cluster Computing
Matei Zaharia, Mosharaf Chowdhury, Tathagata Das, Ankur Dave,
Justin Ma, Murphy McCauley, Michael Franklin, Scott Shenker, Ion Stoica
usenix.org/system/files/conference/nsdi12/nsdi12-final138.pdf
19
20. A Brief History: Functional Programming for Big Data
In addition to simple map and reduce operations,
Spark supports SQL queries, streaming data, and
complex analytics such as machine learning and
graph algorithms out-of-the-box.
Better yet, combine these capabilities seamlessly
into one integrated workflow…
20
21. A Brief History: Key distinctions for Spark vs. MapReduce
• generalized patterns
⇒ unified engine for many use cases
• lazy evaluation of the lineage graph
⇒ reduces wait states, better pipelining
• generational differences in hardware
⇒ off-heap use of large memory spaces
• functional programming / ease of use
⇒ reduction in cost to maintain large apps
• lower overhead for starting jobs
• less expensive shuffles
21
22. TL;DR: Engineering is about costs
Sure, maybe you’ll squeeze slightly better performance
by using many specialized systems…
However, putting on an Eng Director hat, would you
be also prepared to pay the corresponding costs of:
• learning curves for your developers across
several different frameworks
• ops for several different kinds of clusters
• maintenance + troubleshooting mission-critical
apps across several systems
• tech-debt for OSS that ignores the math (80 yrs!)
plus the fundamental h/w trade-offs
22
24. Spark Deconstructed: Log Mining Example
// load error messages from a log into memory!
// then interactively search for various patterns!
// https://gist.github.com/ceteri/8ae5b9509a08c08a1132!
!
// base RDD!
val lines = sc.textFile("hdfs://...")!
!
// transformed RDDs!
val errors = lines.filter(_.startsWith("ERROR"))!
val messages = errors.map(_.split("t")).map(r => r(1))!
messages.cache()!
!
// action 1!
messages.filter(_.contains("mysql")).count()!
!
// action 2!
messages.filter(_.contains("php")).count()
24
25. Driver
Worker
Worker
Worker
Spark Deconstructed: Log Mining Example
We start with Spark running on a cluster…
submitting code to be evaluated on it:
25
26. Spark Deconstructed: Log Mining Example
// base RDD!
val lines = sc.textFile("hdfs://...")!
!
// transformed RDDs!
val errors = lines.filter(_.startsWith("ERROR"))!
val messages = errors.map(_.split("t")).map(r => r(1))!
messages.cache()!
!
// action 1!
messages.filter(_.contains("mysql")).count()!
!
// discussing action 2!
the other part
messages.filter(_.contains("php")).count()
26
27. Spark Deconstructed: Log Mining Example
At this point, take a look at the transformed
RDD operator graph:
scala> messages.toDebugString!
res5: String = !
MappedRDD[4] at map at <console>:16 (3 partitions)!
MappedRDD[3] at map at <console>:16 (3 partitions)!
FilteredRDD[2] at filter at <console>:14 (3 partitions)!
MappedRDD[1] at textFile at <console>:12 (3 partitions)!
HadoopRDD[0] at textFile at <console>:12 (3 partitions)
27
28. Driver
Worker
Worker
Worker
Spark Deconstructed: Log Mining Example
// base RDD!
val lines = sc.textFile("hdfs://...")!
!
// transformed RDDs!
val errors = lines.filter(_.startsWith("ERROR"))!
val messages = errors.map(_.split("t")).map(r => r(1))!
messages.cache()!
!
// action 1!
messages.filter(_.contains("mysql")).count()!
!
// action 2!
medssaigsesc.fuilstesr(i_n.cognt atinhs(e"ph po")t).hcoeuntr() part
28
29. Driver
Worker
Worker
block 1
Worker
block 2
block 3
Spark Deconstructed: Log Mining Example
// base RDD!
val lines = sc.textFile("hdfs://...")!
!
// transformed RDDs!
val errors = lines.filter(_.startsWith("ERROR"))!
val messages = errors.map(_.split("t")).map(r => r(1))!
messages.cache()!
!
// action 1!
messages.filter(_.contains("mysql")).count()!
!
// action 2!
medssaigsesc.fuilstesr(i_n.cognt atinhs(e"ph po")t).hcoeuntr() part
29
30. Driver
Worker
Worker
block 1
Worker
block 2
block 3
Spark Deconstructed: Log Mining Example
// base RDD!
val lines = sc.textFile("hdfs://...")!
!
// transformed RDDs!
val errors = lines.filter(_.startsWith("ERROR"))!
val messages = errors.map(_.split("t")).map(r => r(1))!
messages.cache()!
!
// action 1!
messages.filter(_.contains("mysql")).count()!
!
// action 2!
medssaigsesc.fuilstesr(i_n.cognt atinhs(e"ph po")t).hcoeuntr() part
30
37. Spark Deconstructed:
Looking at the RDD transformations and
actions from another perspective…
action value
RDD
RDD
RDD
// load error messages from a log into memory!
// then interactively search for various patterns!
// https://gist.github.com/ceteri/8ae5b9509a08c08a1132!
!
// base RDD!
val lines = sc.textFile("hdfs://...")!
!
// transformed RDDs!
val errors = lines.filter(_.startsWith("ERROR"))!
val messages = errors.map(_.split("t")).map(r => r(1))!
messages.cache()!
!
// action 1!
messages.filter(_.contains("mysql")).count()!
!
// action 2!
messages.filter(_.contains("php")).count()
transformations RDD
37
42. Unifying the Pieces: Spark SQL
// http://spark.apache.org/docs/latest/sql-programming-guide.html!
!
val sqlContext = new org.apache.spark.sql.SQLContext(sc)!
import sqlContext._!
!
// define the schema using a case class!
case class Person(name: String, age: Int)!
!
// create an RDD of Person objects and register it as a table!
val people = sc.textFile("examples/src/main/resources/
people.txt").map(_.split(",")).map(p => Person(p(0), p(1).trim.toInt))!
!
people.registerAsTempTable("people")!
!
// SQL statements can be run using the SQL methods provided by sqlContext!
val teenagers = sql("SELECT name FROM people WHERE age >= 13 AND age <= 19")!
!
// results of SQL queries are SchemaRDDs and support all the !
// normal RDD operations…!
// columns of a row in the result can be accessed by ordinal!
teenagers.map(t => "Name: " + t(0)).collect().foreach(println)
42
43. Unifying the Pieces: Spark Streaming
// http://spark.apache.org/docs/latest/streaming-programming-guide.html!
!
import org.apache.spark.streaming._!
import org.apache.spark.streaming.StreamingContext._!
!
// create a StreamingContext with a SparkConf configuration!
val ssc = new StreamingContext(sparkConf, Seconds(10))!
!
// create a DStream that will connect to serverIP:serverPort!
val lines = ssc.socketTextStream(serverIP, serverPort)!
!
// split each line into words!
val words = lines.flatMap(_.split(" "))!
!
// count each word in each batch!
val pairs = words.map(word => (word, 1))!
val wordCounts = pairs.reduceByKey(_ + _)!
!
// print a few of the counts to the console!
wordCounts.print()!
!
ssc.start() // start the computation!
ssc.awaitTermination() // wait for the computation to terminate
43
44. MLI: An API for Distributed Machine Learning
Evan Sparks, Ameet Talwalkar, et al.
International Conference on Data Mining (2013)
http://arxiv.org/abs/1310.5426
Unifying the Pieces: MLlib
// http://spark.apache.org/docs/latest/mllib-guide.html!
!
val train_data = // RDD of Vector!
val model = KMeans.train(train_data, k=10)!
!
// evaluate the model!
val test_data = // RDD of Vector!
test_data.map(t => model.predict(t)).collect().foreach(println)!
44
46. Demos, as time permits:
brand new Python support for Streaming in 1.2
github.com/apache/spark/tree/master/examples/src/main/
python/streaming
Twitter Streaming Language Classifier
databricks.gitbooks.io/databricks-spark-reference-applications/
content/twitter_classifier/README.html
!
!
For more Spark learning resources online:
databricks.com/spark-training-resources
46
48. Spark Integrations:
Discover
Insights
Clean Up
Your Data
Run
Sophisticated
Analytics
Integrate With
Many Other
Systems
Use Lots of Different
Data Sources
cloud-based notebooks… ETL… the Hadoop ecosystem…
widespread use of PyData… advanced analytics in streaming…
rich custom search… web apps for data APIs…
low-latency + multi-tenancy…
48
49. Spark Integrations: Unified platform for building Big Data pipelines
Databricks Cloud
databricks.com/blog/2014/07/14/databricks-cloud-making-
big-data-easy.html
youtube.com/watch?v=dJQ5lV5Tldw#t=883
49
54. Spark Integrations: Building data APIs with web apps
Spark + Play
typesafe.com/blog/apache-spark-and-the-typesafe-reactive-
platform-a-match-made-in-heaven
unified compute
web apps
54
55. Spark Integrations: The case for multi-tenancy
Spark + Mesos
spark.apache.org/docs/latest/running-on-mesos.html
+ Mesosphere + Google Cloud Platform
ceteri.blogspot.com/2014/09/spark-atop-mesos-on-google-cloud.html
unified compute
cluster resources
55
57. Spark at Twitter: Evaluation & Lessons Learnt
Sriram Krishnan
slideshare.net/krishflix/seattle-spark-meetup-spark-
at-twitter
• Spark can be more interactive, efficient than MR
• Support for iterative algorithms and caching
• More generic than traditional MapReduce
• Why is Spark faster than Hadoop MapReduce?
• Fewer I/O synchronization barriers
• Less expensive shuffle
• More complex the DAG, greater the
performance improvement
57
Because Use Cases: Twitter
58. Using Spark to Ignite Data Analytics
ebaytechblog.com/2014/05/28/using-spark-to-ignite-
data-analytics/
58
Because Use Cases: eBay/PayPal
59. Hadoop and Spark Join Forces in Yahoo
Andy Feng
spark-summit.org/talk/feng-hadoop-and-spark-join-
forces-at-yahoo/
59
Because Use Cases: Yahoo!
60. Collaborative Filtering with Spark
Chris Johnson
slideshare.net/MrChrisJohnson/collaborative-filtering-
with-spark
• collab filter (ALS) for music recommendation
• Hadoop suffers from I/O overhead
• show a progression of code rewrites, converting
a Hadoop-based app into efficient use of Spark
60
Because Use Cases: Spotify
61. Because Use Cases: Sharethrough
Sharethrough Uses Spark Streaming to
Optimize Bidding in Real Time
Russell Cardullo, Michael Ruggier
2014-03-25
databricks.com/blog/2014/03/25/
sharethrough-and-spark-streaming.html
• the profile of a 24 x 7 streaming app is different than
an hourly batch job…
• take time to validate output against the input…
• confirm that supporting objects are being serialized…
• the output of your Spark Streaming job is only as
reliable as the queue that feeds Spark…
• monoids…
61
62. Because Use Cases: Ooyala
Productionizing a 24/7 Spark Streaming
service on YARN
Issac Buenrostro, Arup Malakar
2014-06-30
spark-summit.org/2014/talk/
productionizing-a-247-spark-streaming-service-
on-yarn
• state-of-the-art ingestion pipeline, processing over
two billion video events a day
• how do you ensure 24/7 availability and fault
tolerance?
• what are the best practices for Spark Streaming and
its integration with Kafka and YARN?
• how do you monitor and instrument the various
62
stages of the pipeline?
63. Because Use Cases: Viadeo
Spark Streaming As Near Realtime ETL
Djamel Zouaoui
2014-09-18
slideshare.net/DjamelZouaoui/spark-streaming
• Spark Streaming is topology-free
• workers and receivers are autonomous and
independent
• paired with Kafka, RabbitMQ
• 8 machines / 120 cores
• use case for recommender system
• issues: how to handle lost data, serialization
63
64. Because Use Cases: Stratio
Stratio Streaming: a new approach to
Spark Streaming
David Morales, Oscar Mendez
2014-06-30
spark-summit.org/2014/talk/stratio-streaming-
a-new-approach-to-spark-streaming
• Stratio Streaming is the union of a real-time
messaging bus with a complex event processing
engine using Spark Streaming
• allows the creation of streams and queries on the fly
• paired with Siddhi CEP engine and Apache Kafka
• added global features to the engine such as auditing
64
and statistics
65. Because Use Cases: Guavus
Guavus Embeds Apache Spark
into its Operational Intelligence Platform
Deployed at the World’s Largest Telcos
Eric Carr
2014-09-25
databricks.com/blog/2014/09/25/guavus-embeds-apache-spark-into-
its-operational-intelligence-platform-deployed-at-the-worlds-
largest-telcos.html
• 4 of 5 top mobile network operators, 3 of 5 top
Internet backbone providers, 80% MSOs in NorAm
• analyzing 50% of US mobile data traffic, +2.5 PB/day
• latency is critical for resolving operational issues
before they cascade: 2.5 MM transactions per second
• “analyze first” not “store first ask questions later”
65
66. Why Spark is the Next Top (Compute) Model
Dean Wampler
slideshare.net/deanwampler/spark-the-next-top-
compute-model
• Hadoop: most algorithms are much harder to
implement in this restrictive map-then-reduce
model
• Spark: fine-grained “combinators” for
composing algorithms
• slide #67, any questions?
66
Because Use Cases: Typesafe
67. Installing the Cassandra / Spark OSS Stack
Al Tobey
tobert.github.io/post/2014-07-15-installing-cassandra-
spark-stack.html
• install+config for Cassandra and Spark together
• spark-cassandra-connector integration
• examples show a Spark shell that can access
tables in Cassandra as RDDs with types pre-mapped
and ready to go
67
Because Use Cases: DataStax
68. One platform for all: real-time, near-real-time,
and offline video analytics on Spark
Davis Shepherd, Xi Liu
spark-summit.org/talk/one-platform-for-all-real-
time-near-real-time-and-offline-video-analytics-
on-spark
68
Because Use Cases: Conviva
70. Approximations
19-20c. statistics emphasized defensibility
in lieu of predictability, based on analytic
variance and goodness-of-fit tests
!
That approach inherently led toward a
manner of computational thinking based
on batch windows
!
They missed a subtle point…
70
71. 21c. shift towards modeling based on probabilistic
approximations: trade bounded errors for greatly
reduced resource costs
highlyscalable.wordpress.com/2012/05/01/
probabilistic-structures-web-analytics-data-
mining/
71
Approximations
72. 21c. shift towards modeling based on probabil
approximations: trade bounded errors for greatly
reduced resource costs
Twitter catch-phrase:
“Hash, don’t sample”
highlyscalable.wordpress.com/2012/05/01/
probabilistic-structures-web-analytics-data-
mining/
72
Approximations
73. a fascinating and relatively new area, pioneered
by relatively few people – e.g., Philippe Flajolet
provides approximation, with error bounds –
in general uses significantly less resources
(RAM, CPU, etc.)
many algorithms can be constructed from
combinations of read and write monoids
aggregate different ranges by composing
hashes, instead of repeating full-queries
73
Approximations
74. algorithm use case example
Count-Min Sketch frequency summaries code
HyperLogLog set cardinality code
Bloom Filter set membership
MinHash
set similarity
DSQ streaming quantiles
SkipList ordered sequence search
74
Approximations
75. algorithm use case example
Count-Min Sketch frequency summaries code
HyperLogLog set cardinality code
suggestion: consider these
as your most quintessential
collections data types at scale
Bloom Filter set membership
MinHash
set similarity
DSQ streaming quantiles
SkipList ordered sequence search
75
Approximations
76. • sketch algorithms: trade bounded errors for
orders of magnitude less required resources,
e.g., fit more complex apps in memory
• multicore + large memory spaces (off heap) are
increasing the resources per node in a cluster
• containers allow for finer-grain allocation of
cluster resources and multi-tenancy
• monoids, etc.: guarantees of associativity within
the code allow for more effective distributed
computing, e.g., partial aggregates
• less resources must be spent sorting/windowing
data prior to working with a data set
• real-time apps, which don’t have the luxury of
anticipating data partitions, can respond quickly
76
Approximations
77. Probabilistic Data Structures for Web
Analytics and Data Mining
Ilya Katsov (2012-05-01)
A collection of links for streaming
algorithms and data structures
Debasish Ghosh
Aggregate Knowledge blog (now Neustar)
Timon Karnezos, Matt Curcio, et al.
Probabilistic Data Structures and
Breaking Down Big Sequence Data
C. Titus Brown, O'Reilly (2010-11-10)
Algebird
Avi Bryant, Oscar Boykin, et al. Twitter (2012)
Mining of Massive Datasets
Jure Leskovec, Anand Rajaraman,
Jeff Ullman, Cambridge (2011)
77
Approximations
79. certification:
Apache Spark developer certificate program
• http://oreilly.com/go/sparkcert
• defined by Spark experts @Databricks
• assessed by O’Reilly Media
• establishes the bar for Spark expertise
79
80. community:
spark.apache.org/community.html
video+slide archives: spark-summit.org
events worldwide: goo.gl/2YqJZK
resources: databricks.com/spark-training-resources
workshops: databricks.com/spark-training
Intro to Spark
Spark
AppDev
Spark
DevOps
Spark
DataSci
Distributed ML
on Spark
Streaming Apps
on Spark
Spark +
Cassandra
80
81. books:
Fast Data Processing
with Spark
Holden Karau
Packt (2013)
shop.oreilly.com/product/
9781782167068.do
Spark in Action
Chris Fregly
Manning (2015*)
sparkinaction.com/
Learning Spark
Holden Karau,
Andy Konwinski,
Matei Zaharia
O’Reilly (2015*)
shop.oreilly.com/product/
0636920028512.do
81
82. events:
Big Data Spain
Madrid, Nov 17-18
bigdataspain.org
Strata EU
Barcelona, Nov 19-21
strataconf.com/strataeu2014
Data Day Texas
Austin, Jan 10
datadaytexas.com
Strata CA
San Jose, Feb 18-20
strataconf.com/strata2015
Spark Summit East
NYC, Mar 18-19
spark-summit.org/east
Strata EU
London, May 5-7
strataconf.com/big-data-conference-uk-2015
Spark Summit 2015
SF, Jun 15-17
spark-summit.org
82
83. presenter:
monthly newsletter for updates,
events, conf summaries, etc.:
liber118.com/pxn/
Just Enough Math
O’Reilly, 2014
justenoughmath.com
preview: youtu.be/TQ58cWgdCpA
Enterprise Data Workflows
with Cascading
O’Reilly, 2013
shop.oreilly.com/product/
0636920028536.do 83