Hadoop & PHP

(PHPNW 17 & Symfony Live London 17)

Transcript

1. HADOOP & PHP PHPNW 2017 @MICHAELCULLUMUK
2. HADOOP & PHP PHPNW 2017 @MICHAELCULLUMUK
3. @MICHAELCULLUMUK ME?
4. MICHAEL CULLUM @MICHAELCULLUMUK
5. @MICHAELCULLUMUK
6. @MICHAELCULLUMUK APACHE HADOOP SUITE OF APPLICATIONS
7. @MICHAELCULLUMUK HADOOP
8. @MICHAELCULLUMUK BIG DATA
9. @MICHAELCULLUMUK
10. @MICHAELCULLUMUK HADOOP 2003: GOOGLE FILE SYSTEM
 2004: MAPREDUCE: SIMPLIFIED DATA PROCESSING 
 ON LARGE CLUSTERS
 2006: HADOOP DEVELOPMENT BEGINS (AND NAMED AFTER A TOY ELEPHANT)
11. @MICHAELCULLUMUK HADOOP 2006: 1.8 TB IN 47.9 HOURS
 2007: 1PB IN 12.13 HOURS, 1.37 TB/MIN
 2008: 1PB IN 6.03 HOURS, 2.76 TB/MIN
 2010: 1PB, 2.95 HOURS, 5.65 TB/MIN
 2011: 1PB, 0.55 HOURS, 30.3 TB/MIN
 2012: 50PB, 23 HOURS, 36.2 TB/MIN
 2014: SPARK ALLOWED - SORT 5 TIMES AS FAST. 10 TIMES FEWER NODES
12. HADOOP COMPONENTS Hadoop Core Hadoop YARN Hadoop HDFS Hadoop Map-Reduce
13. @MICHAELCULLUMUK Hadoop YARN
 Resources Management Hadoop Map-Reduce
 Method of executing large tasks
14. OTHER TOOLS ▸ Hive - Relational style database ▸ Bigtop - Quickly setup a test cluster ▸ Pig - High level programming language for MapReduce jobs ▸ Sqoop - For importing/reading MySQL and other RDBMS ▸ Spark - Alternative to MapReduce designed for fast analytics ▸ Flume - Streaming data collection / aggregation manager ▸ Oozie - MapReduce Workflow Manager & Scheduler ▸ Whirr - Deployment of clusters to AWS ▸ HBase - Low-latency distributed, non- relational database ▸ Zookeeper - Distributed application HA management ▸ HCatalog - Interop between Pig and Hive
15. @MICHAELCULLUMUK ▸ Hadoop HDFS: How data is stored, accessed and distributed under the hood ▸ Hive: Using Hadoop as an RDBMS; writing ▸ Presto: A Facebook library we can use to query the cluster; reading ▸ Phresto: A library to make Presto accessible from PHP userland ▸ Phresto & Doctrine CONTENTS
16. @MICHAELCULLUMUK HADOOP HDFS
17. @MICHAELCULLUMUK ARCHITECTURE Namenode Datanodes
18. @MICHAELCULLUMUK READING A FILE FROM HDFS file.txt 1: 192.168.0.2, 192.168.0.5
 2: 192.168.0.3, 192.168.0.12 Namenode Datanode: 191.168.0.5 Datanode: 192.168.0.3 Read block 1 {content} {content} Read block 2 Client Library
19. @MICHAELCULLUMUK ▸ ✓ Hadoop HDFS: How data is stored, accessed and distributed under the hood ▸ Hive: Using Hadoop as an RDBMS; importing files ▸ Presto: A Facebook library we can use to query the cluster ▸ Phresto: A library to make Presto accessible from PHP userland ▸ Phresto & Doctrine CONTENTS
20. @MICHAELCULLUMUK HIVE
21. @MICHAELCULLUMUK METASTORE
22. @MICHAELCULLUMUK IMPORTING OF FILES
23. @MICHAELCULLUMUK TSV -> ORC
24. IMPORTING TO HIVE IS EASY while read filename; do echo $filename hadoop fs -put /home/michael/for_hadoop_import/$filename /tmp/ echo "LOAD DATA INPATH '/tmp/$filename' INTO TABLE temp_csv; INSERT INTO TABLE temp_orc SELECT * FROM temp_csv; TRUNCATE TABLE temp_csv;" | hive done
25. @MICHAELCULLUMUK LIVE DEMO (IF WIFI WORKS)
26. @MICHAELCULLUMUK ▸ ✓ Hadoop HDFS: How data is stored, accessed and distributed under the hood ▸ ✓ Hive: Using Hadoop as an RDBMS; importing files ▸ Presto: A Facebook library we can use to query the cluster ▸ Phresto: A library to make Presto accessible from PHP userland ▸ Phresto & Doctrine CONTENTS
27. @MICHAELCULLUMUK PRESTO
28. @MICHAELCULLUMUK “DISTRIBUTED SQL ENGINE”
29. @MICHAELCULLUMUK ▸ Hive ▸ MySQL ▸ Cassandra ▸ MongoDB ▸ PostgreSQL ▸ Redis ▸ SQL Server ▸ JMX ▸ REST API ▸ Local files ▸ Memory CONNECTORS
30. @MICHAELCULLUMUK HDF Your (PHP) application Presto Coordinator Presto Workers HDFS Hive Metastore
31. @MICHAELCULLUMUK IT HAS A UI TOO ;)
32. @MICHAELCULLUMUK LIVE DEMO #2
33. @MICHAELCULLUMUK ▸ ✓ Hadoop HDFS: How data is stored, accessed and distributed under the hood ▸ ✓ Hive: Using Hadoop as an RDBMS; importing files ▸ ✓ Presto: A Facebook library we can use to query the cluster ▸ Phresto: A library to make Presto accessible from PHP userland ▸ Phresto & Doctrine CONTENTS
34. @MICHAELCULLUMUK PHRESTO
35. @MICHAELCULLUMUK PHP
36. @MICHAELCULLUMUK REST API
37. @MICHAELCULLUMUK
38. @MICHAELCULLUMUK SIMPLE PHP CLIENT $socket = new \SamKnows\Phresto\Client\RemoteHost('http', ‘coordinator.hostname.com', '8080'); $connection = new \SamKnows\Phresto\Client\HttpConnection($socket, new NullLogger(), 'Michael\'s Macbook', 'Michael', '', 0, 0, 0, false); $result = $connection->executeQuery("SELECT * FROM table WHERE id=5')", 'hive', ‘database_name’); $resultClass = $result->getResult();
39. @MICHAELCULLUMUK TAP INTO PRESTO FUNCTIONALITY TOO $connection->getNodeList(); $connection->getClusterStatus(); $connection->getServerInfo(); $connection->getQueriesStatus(); $connection->getQueryStatus('20170508_163921_00807_v2znp');
40. @MICHAELCULLUMUK SAMKNOWS/PHRESTO
41. @MICHAELCULLUMUK ▸ ✓ Hadoop HDFS: How data is stored, accessed and distributed under the hood ▸ ✓ Hive: Using Hadoop as an RDBMS; importing files ▸ ✓ Presto: A Facebook library we can use to query the cluster ▸ ✓ Phresto: A library to make Presto accessible from PHP userland ▸ Phresto & Doctrine CONTENTS
42. @MICHAELCULLUMUK DOCTRINE & SYMFONY
43. @MICHAELCULLUMUK DOCTRINE DBAL / PDO-STYLE INTERFACE $configuration = new Configuration(); $params = [ 'host' => 'coordinator.hostname.com', 'port' => '8080', 'user' => 'Michael', 'password' => '', 'source' => __FILE__, 'protocol' => 'http', 'catalog' => 'hive', 'schema' => 'database_name', 'driverClass' => Driver::class, ]; $connection = DriverManager::getConnection($params, $configuration); $result = $connection->executeQuery('SELECT * FROM messages LIMIT 1'); $result->fetchAll(); foreach ($result->fetch() as $row) { var_dump($row); }
44. @MICHAELCULLUMUK SYMFONY # Doctrine Configuration doctrine: dbal: default_connection: presto connections: presto: driver_class: '\SamKnows\Phresto\Doctrine\DBAL\Driver\Phresto\Driver' host: 'coordinator.hostname.com' port: '8080' user: 'Michael' password: 'password' charset: UTF8 options: source: 'Michaels MBP' catalog: 'hive' schema: 'table_name' protocol: 'http' Use parameters for these
 configuration values
45. @MICHAELCULLUMUK ▸ ✓ Hadoop HDFS: How data is stored, accessed and distributed under the hood ▸ ✓ Hive: Using Hadoop as an RDBMS; importing files ▸ ✓ Presto: A Facebook library we can use to query the cluster ▸ ✓ Phresto: A library to make Presto accessible from PHP userland ▸ ✓ Phresto & Doctrine CONTENTS
46. @MICHAELCULLUMUK ANY QUESTIONS?
47. THANKS @MICHAELCULLUMUK
48. HADOOP & PHP PHPNW 2017 @MICHAELCULLUMUK