Agile analytics applications on hadoop

Agile Analytics Applications Russell Jurney (@rjurney) - Hadoop Evangelist @Hortonworks Formerly Viz, Data Science at Ning, LinkedIn HBase Dashboards, Career Explorer, InMaps © Hortonworks Inc. 2012 1

Agile Data - The Book (March, 2013) Read it now on OFPS A philosophy, not the only way But still, its good! Really! © Hortonworks Inc. 2012 2

Situation in Brief © Hortonworks Inc. 2012 3

Agile Application Development • LAMP stack mature • Post-Rails frameworks to choose from • We enable rapid feedback and agility + NoSQL © Hortonworks Inc. 2012 4

Data Warehousing © Hortonworks Inc. 2012 5

Scientific Computing / HPC • ‘Smart kid’ only: MPI, Globus, etc. • Math heavy Tubes and Mercury (old school) Cores and Spindles (new school) © Hortonworks Inc. 2012 6

Data Science? Application Data Warehousing Development 33% 33% 33% Scientiﬁc Computing / HPC © Hortonworks Inc. 2012 7

Data Center as Computer • Warehouse Scale Computers and applications “A key challenge for architects of WSCs is to smooth out these discrepancies in a cost efﬁcient manner.” Click here. © Hortonworks Inc. 2012 8

Hadoop to the Rescue! Big data refinery / Modernize ETL Audio, Web, Mobile, CRM, Video, ERP, SCM, … Images New Data Business Transactions Docs, Sources Text, & Interactions XML HDFS Web Logs, Clicks Big Data Social, Refinery SQL NoSQL NewSQL Graph, Feeds ETL EDW MPP NewSQL Sensors, Devices, RFID Business Spatial, GPS Apache Hadoop Intelligence & Analytics Events, Other Dashboards, Reports, Visualization, … Page 7 © Hortonworks Inc. 2012 I stole this slide from Eric. 9

Hadoop to the Rescue! • A department can afford a Hadoop cluster, let alone an org • Dump all your data in one place: HDFS • JOIN like crazy! • ETL like whoah! • An army of mappers and reducers at your command • Now what? © Hortonworks Inc. 2012 10

Analytics Apps: It takes a Team • Broad skill-set to make useful apps • Basically nobody has them all • Application development is inherently collaborative © Hortonworks Inc. 2012 11

How to get insight into product? • Back-end has gotten t-h-i-c-k-e-r • Generating $$$ insight can take 10-100x app dev • Timeline disjoint: analytics vs agile app-dev/design • How do you ship insights efficiently? • How do you collaborate on research vs developer timeline? © Hortonworks Inc. 2012 12

The Wrong Way - Part One “We made a great design. Your job is to predict the future for it.” © Hortonworks Inc. 2012 13

The Wrong Way - Part Two “Whats taking you so long to reliably predict the future?” © Hortonworks Inc. 2012 14

The Wrong Way - Part Three “The users don’t understand what 86% true means.” © Hortonworks Inc. 2012 15

The Wrong Way - Part Four GHJIAEHGIEhjagigehganbanbigaebjnain!!!!!RJ(@J?!! © Hortonworks Inc. 2012 16

The Wrong Way - Inevitable Conclusion Plane Mountain © Hortonworks Inc. 2012 17

Reminds me of... © Hortonworks Inc. 2012 18

Chief Problem You can’t design insight in analytics applications. You discover it. You discover by exploring. © Hortonworks Inc. 2012 19

-> Strategy So make an app for exploring your data. Iterate and publish intermediate results. Which becomes a palette for what you ship. © Hortonworks Inc. 2012 20

Data Design • Not the 1st query that = insight, its the 15th, or the 150th • Capturing “Ah ha!” moments • Slow to do those in batch... • Faster, better context in an interactive web application. • Pre-designed charts wind up terrible. So bad. • Easy to invest man-years in the wrong statistical models • Semantics of presenting predictions are complex, delicate • Opportunity lies at intersection of data & design © Hortonworks Inc. 2012 21

How do we get back to Agile? © Hortonworks Inc. 2012 22

Statement of Principles (then tricks, with code) © Hortonworks Inc. 2012 23

Setup an environment where... • Insights repeatedly produced • Iterative work shared with entire team • Interactive from day 0 • Data model is consistent end-to-end • Minimal impedance between layers • Scope and depth of insights grow • Insights form the palette for what you ship • Until the application pays for itself and more © Hortonworks Inc. 2012 24

Value document > relation Most data is dirty. Most data is semi-structured or un-structured. Rejoice! © Hortonworks Inc. 2012 25

Value document > relation Note: Hive/ArrayQL/NewSQL’s support of documents/array types blur this distinction. © Hortonworks Inc. 2012 26

Value imperative > declarative • We don’t know what we want to SELECT. • Data is dirty - check each step, clean iteratively. • 85% of data scientist’s time spent munging. See: ETL. • Imperative is optimized for our process. • Process = iterative, snowballing insight • Efficiency matters, self optimize © Hortonworks Inc. 2012 27

Value dataflow > SELECT © Hortonworks Inc. 2012 28

Example dataflow: ETL + email sent count © Hortonworks Inc. 2012 (I can’t read this either. Get a big version here.) 29

Value Pig > Hive (for app-dev) • Pigs eat ANYTHING • Pig is optimized for refining data, as opposed to consuming it • Pig is imperative, iterative • Pig is dataflows, and SQLish (but not SQL) • Code modularization/re-use: Pig Macros • ILLUSTRATE speeds dev time (even UDFs) • Easy UDFs in Java, JRuby, Jython, Javascript • Pig Streaming = use any tool, period. • Easily prepare our data as it will appear in our app. • If you prefer Hive, use Hive. But actually, I wish Pig and Hive were one tool. Pig, then Hive, then Pig, then Hive... See: HCatalog for Pig/Hive integration, and this post. © Hortonworks Inc. 2012 30

Localhost vs Petabyte scale: same tools • Simplicity essential to scalability: highest level tools we can • Prepare a good sample - tricky with joins, easy with documents • Local mode: pig -l /tmp -x local -v -w • Frequent use of ILLUSTRATE • 1st: Iterate, debug & publish locally • 2nd: Run on cluster, publish to team/customer • Consider skipping Object-Relational-Mapping (ORM) • We do not trust ‘databases,’ only HDFS @ n=3. • Everything we serve in our app is re-creatable via Hadoop. © Hortonworks Inc. 2012 31

Data-Value Pyramid Climb it. Do not skip steps. See here. © Hortonworks Inc. 2012 32

0/1) Display atomic records on the web © Hortonworks Inc. 2012 33

0.0) Document-serialize events • Protobuf • Thrift • JSON • Avro - I use Avro because the schema is onboard. © Hortonworks Inc. 2012 34

0.1) Documents via Relation ETL enron_messages = load '/enron/enron_messages.tsv' as ( message_id:chararray, sql_date:chararray, from_address:chararray, from_name:chararray, subject:chararray, body:chararray ); enron_recipients = load '/enron/enron_recipients.tsv' as ( message_id:chararray, reciptype:chararray, address:chararray, name:chararray); split enron_recipients into tos IF reciptype=='to', ccs IF reciptype=='cc', bccs IF reciptype=='bcc'; headers = cogroup tos by message_id, ccs by message_id, bccs by message_id parallel 10; with_headers = join headers by group, enron_messages by message_id parallel 10; emails = foreach with_headers generate enron_messages::message_id as message_id, CustomFormatToISO(enron_messages::sql_date, 'yyyy-MM-dd HH:mm:ss') as date, TOTUPLE(enron_messages::from_address, enron_messages::from_name) as from:tuple(address:chararray, name:chararray), enron_messages::subject as subject, enron_messages::body as body, headers::tos.(address, name) as tos, headers::ccs.(address, name) as ccs, headers::bccs.(address, name) as bccs; store emails into '/enron/emails.avro' using AvroStorage( Example here. © Hortonworks Inc. 2012 35

0.2) Serialize events from streams class GmailSlurper(object): ... def init_imap(self, username, password): self.username = username self.password = password try: imap.shutdown() except: pass self.imap = imaplib.IMAP4_SSL('imap.gmail.com', 993) self.imap.login(username, password) self.imap.is_readonly = True ... def write(self, record): self.avro_writer.append(record) ... def slurp(self): if(self.imap and self.imap_folder): for email_id in self.id_list: (status, email_hash, charset) = self.fetch_email(email_id) if(status == 'OK' and charset and 'thread_id' in email_hash and 'froms' in email_hash): print email_id, charset, email_hash['thread_id'] self.write(email_hash) © Hortonworks Inc. 2012 Scrape your own gmail in Python and Ruby. 36

0.3) ETL Logs log_data = LOAD 'access_log' USING org.apache.pig.piggybank.storage.apachelog.CommongLogLoader AS (remoteAddr, remoteLogname, user, time, method, uri, proto, bytes); © Hortonworks Inc. 2012 37

1) Plumb atomic events -> browser (Example stack that enables high productivity) © Hortonworks Inc. 2012 38

Lots of Stack Options with Examples • Pig with Voldemort, Ruby, Sinatra: example • Pig with ElasticSearch: example • Pig with MongoDB, Node.js: example • Pig with Cassandra, Python Streaming, Flask: example • Pig with HBase, JRuby, Sinatra: example • Pig with Hive via HCatalog: example (trivial on HDP) • Up next: Accumulo, Redis, MySQL, etc. © Hortonworks Inc. 2012 39

1.1) cat our Avro serialized events me$ cat_avro ~/Data/enron.avro { u'bccs': [], u'body': u'scamming people, blah blah', u'ccs': [], u'date': u'2000-08-28T01:50:00.000Z', u'from': {u'address': u'bob.dobbs@enron.com', u'name': None}, u'message_id': u'<1731.10095812390082.JavaMail.evans@thyme>', u'subject': u'Re: Enron trade for frop futures', u'tos': [ {u'address': u'connie@enron.com', u'name': None} ] } © Hortonworks Inc. 2012 Get cat_avro in python, ruby 40

1.2) Load our events in Pig me$ pig -l /tmp -x local -v -w grunt> enron_emails = LOAD '/enron/emails.avro' USING AvroStorage(); grunt> describe enron_emails emails: { message_id: chararray, datetime: chararray, from:tuple(address:chararray,name:chararray) subject: chararray, body: chararray, tos: {to: (address: chararray,name: chararray)}, ccs: {cc: (address: chararray,name: chararray)}, bccs: {bcc: (address: chararray,name: chararray)} } © Hortonworks Inc. 2012 41

1.3) ILLUSTRATE our events in Pig grunt> illustrate enron_emails --------------------------------------------------------------------------- | emails | | message_id:chararray | | datetime:chararray | | from:tuple(address:chararray,name:chararray) | | subject:chararray | | body:chararray | | tos:bag{to:tuple(address:chararray,name:chararray)} | | ccs:bag{cc:tuple(address:chararray,name:chararray)} | | bccs:bag{bcc:tuple(address:chararray,name:chararray)} | --------------------------------------------------------------------------- | | | <1731.10095812390082.JavaMail.evans@thyme> | | 2001-01-09T06:38:00.000Z | | (bob.dobbs@enron.com, J.R. Bob Dobbs) | | Re: Enron trade for frop futures | | scamming people, blah blah | | {(connie@enron.com,)} | | {} | | {} | Upgrade to Pig 0.10+ © Hortonworks Inc. 2012 42

1.4) Publish our events to a ‘database’ From Avro to MongoDB in one command: pig -l /tmp -x local -v -w -param avros=enron.avro -param mongourl='mongodb://localhost/enron.emails' avro_to_mongo.pig Which does this: /* MongoDB libraries and configuration */ register /me/mongo-hadoop/mongo-2.7.3.jar register /me/mongo-hadoop/core/target/mongo-hadoop-core-1.1.0-SNAPSHOT.jar register /me/mongo-hadoop/pig/target/mongo-hadoop-pig-1.1.0-SNAPSHOT.jar /* Set speculative execution off to avoid chance of duplicate records in Mongo */ set mapred.map.tasks.speculative.execution false set mapred.reduce.tasks.speculative.execution false define MongoStorage com.mongodb.hadoop.pig.MongoStorage(); /* Shortcut */ /* By default, lets have 5 reducers */ set default_parallel 5 avros = load '$avros' using AvroStorage(); store avros into '$mongourl' using MongoStorage(); © Hortonworks Inc. 2012 Full instructions here. 43

1.5) Check events in our ‘database’ $ mongo enron MongoDB shell version: 2.0.2 connecting to: enron > show collections emails system.indexes > db.emails.findOne({message_id: "<1731.10095812390082.JavaMail.evans@thyme>"}) { " "_id" : ObjectId("502b4ae703643a6a49c8d180"), " "message_id" : "<1731.10095812390082.JavaMail.evans@thyme>", " "date" : "2001-01-09T06:38:00.000Z", " "from" : { "address" : "bob.dobbs@enron.com", "name" : "J.R. Bob Dobbs" }, " "subject" : Re: Enron trade for frop futures, " "body" : "Scamming more people...", " "tos" : [ { "address" : "connie@enron", "name" : null } ], " "ccs" : [ ], " "bccs" : [ ] } © Hortonworks Inc. 2012 44

1.6) Publish events on the web require 'rubygems' require 'sinatra' require 'mongo' require 'json' connection = Mongo::Connection.new database = connection['agile_data'] collection = database['emails'] get '/email/:message_id' do |message_id| data = collection.find_one({:message_id => message_id}) JSON.generate(data) end © Hortonworks Inc. 2012 45

1.6) Publish events on the web © Hortonworks Inc. 2012 46

Whats the point? • A designer can work against real data. • An application developer can work against real data. • A product manager can think in terms of real data. • Entire team is grounded in reality! • You’ll see how ugly your data really is. • You’ll see how much work you have yet to do. • Ship early and often! • Feels agile, don’t it? Keep it up! © Hortonworks Inc. 2012 47

1.7) Wrap events with Bootstrap <link href="/static/bootstrap/docs/assets/css/bootstrap.css" rel="stylesheet"> </head> <body> <div class="container" style="margin-top: 100px;"> <table class="table table-striped table-bordered table-condensed"> <thead> {% for key in data['keys'] %} <th>{{ key }}</th> {% endfor %} </thead> <tbody> <tr> {% for value in data['values'] %} <td>{{ value }}</td> {% endfor %} </tr> </tbody> </table> </div> </body> Complete example here with code here. © Hortonworks Inc. 2012 48

1.7) Wrap events with Bootstrap © Hortonworks Inc. 2012 49

Refine. Add links between documents. Not the Mona Lisa, but coming along... See: here © Hortonworks Inc. 2012 50

The Mona Lisa. In pure CSS. © Hortonworks Inc. 2012 See: here 51

1.8) List links to sorted events Use Pig, serve/cache a bag/array of email documents: pig -l /tmp -x local -v -w emails_per_user = foreach (group emails by from.address) { sorted = order emails by date; last_1000 = limit sorted 1000; generate group as from_address, emails as emails; }; store emails_per_user into '$mongourl' using MongoStorage(); Use your ‘database’, if it can sort. mongo enron > db.emails.ensureIndex({message_id: 1}) > db.emails.find().sort({date:0}).limit(10).pretty() { { " "_id" : ObjectId("4f7a5da2414e4dd0645d1176"), " "message_id" : "<CA+bvURyn-rLcH_JXeuzhyq8T9RNq+YJ_Hkvhnrpk8zfYshL-wA@mail.gmail.com>", " "from" : [ ... © Hortonworks Inc. 2012 52

1.8) List links to sorted documents © Hortonworks Inc. 2012 53

1.9) Make it searchable... If you have list, search is easy with ElasticSearch and Wonderdog... /* Load ElasticSearch integration */ register '/me/wonderdog/target/wonderdog-1.0-SNAPSHOT.jar'; register '/me/elasticsearch-0.18.6/lib/*'; define ElasticSearch com.infochimps.elasticsearch.pig.ElasticSearchStorage(); emails = load '/me/tmp/emails' using AvroStorage(); store emails into 'es://email/email?json=false&size=1000' using ElasticSearch('/me/ elasticsearch-0.18.6/config/elasticsearch.yml', '/me/elasticsearch-0.18.6/plugins'); Test it with curl: curl -XGET 'http://localhost:9200/email/email/_search?q=hadoop&pretty=true&size=1' ElasticSearch has no security features. Take note. Isolate. © Hortonworks Inc. 2012 54

From now on we speed up... Don’t worry, its in the book and on the blog. http://hortonworks.com/blog/ © Hortonworks Inc. 2012 55

2) Create Simple Charts © Hortonworks Inc. 2012 56

2) Create Simple Charts • Start with an HTML table on general principle. • Then use nvd3.js - reusable charts for d3.js • Aggregate by properties & displaying is first step in entity resolution • Start extracting entities. Ex: people, places, topics, time series • Group documents by entities, rank and count. • Publish top N, time series, etc. • Fill a page with charts. • Add a chart to your event page. © Hortonworks Inc. 2012 58

2.1) Top N (of anything) in Pig pig -l /tmp -x local -v -w top_things = foreach (group things by key) { sorted = order things by arbitrary_rank desc; top_10_things = limit sorted 10; generate group as key, top_10_things as top_10_things; }; store top_n into '$mongourl' using MongoStorage(); Remember, this is the same structure the browser gets as json. This would make a good Pig Macro. © Hortonworks Inc. 2012 59

2.2) Time Series (of anything) in Pig pig -l /tmp -x local -v -w /* Group by our key and date rounded to the month, get a total */ things_by_month = foreach (group things by (key, ISOToMonth(datetime)) generate flatten(group) as (key, month), COUNT_STAR(things) as total; /* Sort our totals per key by month to get a time series */ things_timeseries = foreach (group things_by_month by key) { timeseries = order things by month; generate group as key, timeseries as timeseries; }; store things_timeseries into '$mongourl' using MongoStorage(); Yet another good Pig Macro. © Hortonworks Inc. 2012 60

Data processing in our stack A new feature in our application might begin at any layer... great! omghi2u! I’m creative! I’m creative too! where r my legs? I know Pig! I <3 Javascript! send halp Any team member can add new features, no problemo! © Hortonworks Inc. 2012 61

Data processing in our stack ... but we shift the data-processing towards batch, as we are able. See real example here. Ex: Overall total emails calculated in each layer © Hortonworks Inc. 2012 62

3.0) From charts to reports... • Extract entities from properties we aggregated by in charts (Step 2) • Each entity gets its own type of web page • Each unique entity gets its own web page • Link to entities as they appear in atomic event documents (Step 1) • Link most related entities together, same and between types. • More visualizations! • Parametize results via forms. © Hortonworks Inc. 2012 65

3.3) Get people clicking. Learn. • Explore this web of generated pages, charts and links! • Everyone on the team gets to know your data. • Keep trying out different charts, metrics, entities, links. • See whats interesting. • Figure out what data needs cleaning and clean it. • Start thinking about predictions & recommendations. ‘People’ could be just your team, if data is sensitive. © Hortonworks Inc. 2012 68

4.0) Preparation • We’ve already extracted entities, their properties and relationships • Our charts show where our signal is rich • We’ve cleaned our data to make it presentable • The entire team has an intuitive understanding of the data • They got that understanding by exploring the data • We are all on the same page! © Hortonworks Inc. 2012 70

4.2) Think in different perspectives • Networks • Time Series • Distributions • Natural Language • Probability / Bayes © Hortonworks Inc. 2012 See here. 72

4.3) Sink more time in deeper analysis TF-IDF import 'tfidf.macro'; my_tf_idf_scores = tf_idf(id_body, 'message_id', 'body'); /* Get the top 10 Tf*Idf scores per message */ per_message_cassandra = foreach (group tfidf_all by message_id) { sorted = order tfidf_all by value desc; top_10_topics = limit sorted 10; generate group, top_10_topics.(score, value); } Probability / Bayes sent_replies = join sent_counts by (from, to), reply_counts by (from, to); reply_ratios = foreach sent_replies generate sent_counts::from as from, sent_counts::to as to, (float)reply_counts::total/(float)sent_counts::tot as ratio; reply_ratios = foreach reply_ratios generate from, to, (ratio > 1.0 ? 1.0 : ratio) as ratio; © Hortonworks Inc. 2012 Example with code here and macro here. 73

Example: Packetpig and PacketLoop snort_alerts = LOAD '$pcap' USING com.packetloop.packetpig.loaders.pcap.detection.SnortLoader('$snortconﬁg'); countries = FOREACH snort_alerts GENERATE com.packetloop.packetpig.udf.geoip.Country(src) as country, priority; countries = GROUP countries BY country; countries = FOREACH countries GENERATE group, AVG(countries.priority) as average_severity; STORE countries into 'output/choropleth_countries' using PigStorage(','); Code here. © Hortonworks Inc. 2012 76

Hortonworks Data Platform • Simplify deployment to get started quickly and easily • Monitor, manage any size cluster with familiar console and tools • Only platform to include data 1 integration services to interact with any data • Metadata services opens the platform for integration with existing applications • Dependable high availability architecture  Reduce risks and cost of adoption • Tested at scale to future proof  Lower the total cost to administer and provision your cluster growth  Integrate with your existing ecosystem © Hortonworks Inc. 2012 78

Hortonworks Training The expert source for Apache Hadoop training & certification Role-based Developer and Administration training – Coursework built and maintained by the core Apache Hadoop development team. – The “right” course, with the most extensive and realistic hands-on materials – Provide an immersive experience into real-world Hadoop scenarios – Public and Private courses available Comprehensive Apache Hadoop Certification – Become a trusted and valuable Apache Hadoop expert © Hortonworks Inc. 2012 79

Next Steps? 1 Download Hortonworks Data Platform hortonworks.com/download 2 Use the getting started guide hortonworks.com/get-started 3 Learn more… get support Hortonworks Support • Expert role based training • Full lifecycle technical support • Course for admins, developers across four service levels and operators • Delivered by Apache Hadoop • Certification program Experts/Committers • Custom onsite options • Forward-compatible hortonworks.com/training hortonworks.com/support © Hortonworks Inc. 2012 80

Thank You! Questions & Answers Slides: http://slidesha.re/O8kjaF Follow: @hortonworks and @rjurney Read: hortonworks.com/blog © Hortonworks Inc. 2012 81

Agile analytics applications on hadoop

More Related Content

What's hot

Similar to Agile analytics applications on hadoop

More from Hortonworks

Recently uploaded

Agile analytics applications on hadoop

Editor's Notes