Hive

What is Hive? Data warehousing layer on top of Hadoop – table abstractions SQL-like language (HiveQL) for “batch” data processing SQL is translated a one or series of MapReduce executions Good for ad-hoc reporting queries on HDFS data – however generated MR executions can be sub optimal

What is not… Not a relational database – not transactions – no row updates – no indexes No interactive querying No auditing No accounts

Hive overview Metadata (table definitions, data locations Metadata (table definitions, data locations Hive Engine (compiler, optimizer, executer) Hive Engine (compiler, optimizer, executer) HDFS Hadoop Distributed File System YARN Cluster resource manager MapReduce Lookup s Direct data reading Data processing

Metastore Contains tables definitions and data locations Stored in additional RDBMS – very often on one of a cluster machines – MySQL, PostgreSQL, Oracle, Derby… Used by many other Hadoop components – via HCatalog service

Hive Table Table – Definitions stored in a Hive metastore (RDBMS) – Data are stored in files on HDFS Table definition and data are not tightly coupled Data files can be in various formats – but a type has to by unique within a single table Table partitioning and bucketing is supported EXTERNAL tables (DMLs are not possible)

Interacting with Hive Remotely – JDBC and ODBC drivers – Thrift API Locally – hive-shell (deprecated) – beeline (via JDBC, HiveServer2 support)

Hadoop Cluster Machine Hive Architecture JDBC ODBC Thrift API Hadoop Cluster machine HiveServer2 Thrift Service Hive Driver Metastore Server Thrift Service Metadata RDBMS (table definitions, data locations Metadata RDBMS (table definitions, data locations DB driver Compilation Optimization Execution Cluster Node HDFS MapReduce tasks Cluster Node HDFS MapReduce tasks Cluster Node HDFS MapReduce tasks Cluster Node HDFS MapReduce tasks SQL master node HDFS MapReduce task YARN Request Lookup s

Operations Based on SQL-92 specification DDL – CREATE TABLE, ALTER TABLE, DROP TABLE…. DML – INSERT, INSERT OVERWRITE… SQL – SELECT… DISTINCT…JOIN WHERE…GROUP BY…HAVING…ORDER BY…LIMIT – REGEXP supported – Subqueries only in the FROM cluase

Data Types TINYINT – 1 byte BOOLEAN SMALLINT – 2 bytes INT – 4 bytes BIGINT – 8 bytes DOUBLE STRING STRUCT – named fields MAP – key-value pairs collection ARRAY – order collection of records in the same type

Other features Views Build in functions – floor, rand, cast, case, if, concat, substr etc – ‘show functions’ User defined functions – have to be delivered in jar

MR hands on (1) The problem – Q: „What follows two rainy days in the Geneva region?” – A: „Monday” The goal – Proof if the theory is true or false Solution – Lets take meteo data from GVA and build a histogram of bad weather days of a week Mon | Tue |Wed |Thu | Fr | Sat | Sun Bad weather days count ?

MR hands on (2) – The source data ( – Source: Last 3 years of weather data taken at GVA airport – CSV format What is a bad weather day?: – Weather anomalies between 8am and 10pm "Local time in Geneva (airport)";"T";"P0";"P";"U";"DD";"Ff";"ff10";"WW";"W'W'";"c";"VV";"Td"; " :50";"18.0";"730.4";"767.3";"100";"variable wind direction";"2";"";"";"";"No Significant Clouds";"10.0 and more";"18.0"; " :20";"18.0";"730.4";"767.3";"94";"variable wind direction";"1";"";"";"";"Few clouds (10-30%) 300 m, scattered clouds (40-50%) 3300 m";"10.0 and more";"17.0"; " :50";"19.0";"730.5";"767.3";"88";"Wind blowing from the west";"2";"";"";"";"Few clouds (10-30%) 300 m, broken clouds (60-90%) 5400 m";"10.0 and more";"17.0"; " :20";"19.0";"729.9";"766.6";"83";"Wind blowing from the south-east";"4";"";"";"";"Few clouds (10-30%) 300 m, scattered clouds (40-50%) 2400 m, overcast (100%) 4500 m";"10.0 and more";"16.0"; " :50";"19.0";"729.9";"766.6";"94";"Wind blowing from the east-northeast";"5";"";"Light shower(s), rain";"";"Few clouds (10-30%) 1800 m, scattered clouds (40-50%) 2400 m, broken clouds (60-90%) 3000 m";"10.0 and more";"18.0"; " :20";"20.0";"730.7";"767.3";"88";"Wind blowing from the north-west";"2";"";"Light shower(s), rain, in the vicinity thunderstorm";"";"Few clouds (10-30%) 1800 m, cumulonimbus clouds, broken clouds (60-90%) 2400 m";"10.0 and more";"18.0"; " :50";"22.0";"730.2";"766.6";"73";"Wind blowing from the south";"7";"";"Thunderstorm";"";"Few clouds (10-30%) 1800 m, cumulonimbus clouds, scattered clouds (40-50%) 2100 m, broken clouds (60-90%) 3000 m";"10.0 and more";"17.0"; " :20";"23.0";"729.6";"765.8";"78";"Wind blowing from the west-southwest";"4";"";"Light shower(s), rain, in the vicinity thunderstorm";"";"Few clouds (10-30%) 1740 m, cumulonimbus clouds, scattered clouds (40-50%) 2100 m, broken clouds (60-90%) 3000 m";"10.0 and more";"19.0"; " :50";"23.0";"728.8";"765.0";"65";"variable wind direction";"2";"";"In the vicinity thunderstorm";"";"Scattered clouds (40-50%) 1950 m, cumulonimbus clouds, scattered clouds (40-50%) 2100 m, broken clouds (60-90%) 3300 m";"10.0 and more";"16.0"; " :20";"23.0";"728.2";"764.3";"74";"Wind blowing from the west-northwest";"4";"";"Light thunderstorm, rain";"";"Scattered clouds (40-50%) 1950 m, cumulonimbus clouds, scattered clouds (40-50%) 2100 m, broken clouds (60-90%) 3300 m";"10.0 and more";"18.0"; " :50";"28.0";"728.0";"763.5";"45";"Wind blowing from the south-west";"5";"11";"Thunderstorm";"";"Scattered clouds (40-50%) 1950 m, cumulonimbus clouds, scattered clouds (40-50%) 2100 m, broken clouds (60-90%) 6300 m";"10.0 and more";"15.0"; " :20";"28.0";"728.0";"763.5";"42";"Wind blowing from the north-northeast";"2";"";"In the vicinity thunderstorm";"";"Few clouds (10-30%) 1950 m, cumulonimbus clouds, broken clouds (60-90%) 6300 m";"10.0 and more";"14.0";

MR hands on (3) Designing MapReduce flow " :50";"18.0";... " :20";"18.0„;... " :50";"19.0";... Input Data Input Data Grouping: 1)sum(Value) by key Emiting: Reduce Intermediate output Map Filtering: 1) 8 < HH < 22 2) col9 != „” ; Emiting: Filtering: 1) 8 < HH < 22 2) col9 != „” ; Emiting: Filtering: 1) 8 < HH < 22 2) col9 != „” ; Emiting: Result Result (final) Filtering: 1) 8 < HH < 22 2) col9 != „” ; Emiting: Filtering: 1) 8 < HH < 22 2) col9 != „” ; Emiting: Transforming: 1) Date => day of a week Emiting: Map Intermediate output Grouping: 1)sum(Value) by key Emiting: Reduce

SQL query for the problem select from_unixtime(unix_timestamp(a.day,'yyyy-MM-dd'),'u'), count(*) from ( select from_unixtime( unix_timestamp(time,’dd.MM.yyyy HH:mm'),'yyyy-MM-dd') as day, count(*) as w from data where weather!=‘’ and hour( from_unixtime(unix_timestamp(time,’dd.MM.yyyy HH:mm'))) between 8 and 22 group by from_unixtime(unix_timestamp(time,’dd.MM.yyyy HH:mm'),'yyyy-MM-dd') having count(*)>=1 ) a group by from_unixtime(unix_timestamp(a.day,'yyyy-MM-dd'),'u');

Using Hive CLIs Starting hive shell (deprecated) Use beeline instead (supports new HiveServer2) – Connection in remote mode – Connection in embedded mode > hive >beeline !connect jdbc:hive2://localhost:10000/default !connect jdbc:hive2://

Useful Hive commands Get all databases Set a default database Show tables in a database Show table definition Explaining plan show databases use show tables desc EXPLAIN [EXTENDED|DEPENDENCY|AUTHORIZATION]

Hands on Hive (1) All scripts are available with: To execute a script in beeline Querying the a table – query_.sql Creation of an external table (name=geneva) – external.sql Creation of a external table without “ (name=geneva_clean) – external_clean.sql Creation of a local table (name=weather) – standard.sql wget unzip hive.zip; cd hive !run

Hands-on Hive (2) Creation of a partitioned table (name=weather_part) – partitioned.sql Creation of partitioned and bucketed table (name=weather_buck) – bucketing.sql Creation of a table stored in a parquet format (name=weather_parquet) – parquet.sql Creation of a compressed table (name=weather_compr) – compressed.sql Explain plan – explain.sql Table statistics – stats.sql

Hands on Hive (JDBC) Compile the code Run – set classpath – Execute javac HiveJdbcClient.java java -cp $CLASSPATH HiveJdbcClient source./setHiveEnv.sh

This talk does not include… SerDe – Serializer and Deserializer of data – There are predefined for common data formats – Custom Ser/De can be written by a user Writing UDF Querying Hbase

Summary Provides table abstraction layer on HDFS data SQL on Hadoop translated to MapReduce jobs The Hive Query Language has some limitations For batch processing, not interactive Can append data – But not row updates or deletions Uses external metastore for keeping metadata – The store is used by most of SQL on Hadoop technologies