Apache Tajo - An open source big data warehouse

1. Apache Tajo: An Open Source Big Data Warehouse (what’s new in recent releases) HadoopSphere -‐ Virtual Conclave 2015 Hyunsik Choi, Gruter Inc. (hschoi @ gruter.com) 1

2. Agenda •  Tajo Overview •  Milestones and 0.10 Features •  What’s Next 2

3. Tajo: A Big Data Warehouse System •  Apache Top-‐level project •  Distributed and scalable data warehouse system on various data sources (e.g, HDFS, S3, Hbase, …) •  Low latency, and long running batch queries in a single system •  Features •  ANSI SQL compliance •  Mature SQL features •  ParYYoned table support •  Java/Python UDF support •  JDBC driver and Java-‐based asynchronous API •  Read/Write support of CSV, JSON, RCFile, SequenceFile, Parquet, ORC 3

7. Master

8. Server

17. TajoMaster

23. Slave

24. Server

45. TajoWorker

56. QueryMaster

57. Local

58. Query

59. Engine

60. StorageManager

61. HDFS

62. HBase

65. Client

71. JDBC

72. TSql

73. Web

74. UI

80. Slave

81. Server

102. TajoWorker

113. QueryMaster

114. Local

115. Query

116. Engine

117. StorageManager

123. Slave

124. Server

145. TajoWorker

156. QueryMaster

157. Local

158. Query

159. Engine

160. StorageManager

161. CatalogStore

162. DBMS

163. HCatalog

164. Submit

166. a

167. query

168. Manage

169. metadata

170. Allocate

171. a

172. query

173. send

174. tasks

176. monitor

178. send

179. tasks

181. monitor

183. Tajo Overall Architecture HDFS

184. HBase

185. HDFS

186. HBase

187. 4

188. Common Scenarios •  ExtracYon, TransformaYon, Loading (ETL) •  InteracYve BI/analyYcs on web-‐scale big data •  Data discovery/Exploratory analysis with R and exisYng SQL tools 5

189. Use Cases: Replacement of Commercial DW •  Example: Biggest Telco Company in South Korea •  Goal: •  Replacement of slow ETL workloads on several TB datasets •  Lots daily reports generaYon about users’ behaviors •  Ad-‐hoc analysis on Terabytes data sets •  Key Benefits of Tajo: •  SimplificaYon of DW ETL, OLAP, and Hadoop ETL into an unified system •  Saved license over commercial DW •  Much less cost, more data analysis within the same SLA 6

190. Use Cases: Data Discovery •  Example: Music streaming service (26 million users) •  Goal: •  Analysis on purchase history for target markeYng •  Beneﬁts: •  Query interacYvity on large data sets •  Ability to use exisYng BI visualizaYon tools 7

191. When Tajo is right choice? •  You want an uniﬁed system for batch and interacYve queries on Hadoop, Amazon S3, or Hbase. •  You want a mixed use of Hadoop-‐based DW and RDBMS-‐based DW or want to replace exisYng RDBMS DW. •  You want to use exisYng SQL tools on Hadoop DW 8

192. Milestones 0.8 0.9 0.10 0.11 More features SQL compaYbility Stability AnalyYcal funcYon Eco-‐system expansion More features •  Python UDF •  Nested Schema •  Tablespace support •  Query federaYon •  Beker query scheduler 9

193. Selected Features in 0.10 10

194. Hbase Storage Support •  You can use SQL to access Hbase tables. •  Tajo supports Hbase storage •  CREATE (EXTERNAL)/DROP/INSERT (OVERWRITE)/ SELECT •  Bulk InserYon through Direct HFile wriYng CREATE TABLE hbase_t1 (key TEXT, col1 TEXT, col2 INT) USING hbase WITH ( ‘table’ = ‘t1’, ‘columns’ = ‘:key,cf1:col1,cf2:col2`, ‘hbase.zookeeper.quorum’ = ‘host1:2181,host2:2181’ ) 11

195. BeNer AWS support •  OpYmized for S3 and EMR environments •  Fixed many bugs related to S3 •  EMR bootstrap supported in AWS Labs Github repo •  A quick guide for Tajo on EMR •  hkp://www.gruter.com/blog/semng-‐up-‐a-‐tajo-‐cluster-‐on-‐amazon-‐emr/ •  EMR bootstrap for Tajo on EMR •  hkps://github.com/awslabs/emr-‐bootstrap-‐acYons/tree/master/tajo 12

196. Tajo JDBC Tajo Cluster ETL Tools BI Tools Repor.ng tools BeNer SQL tool support via thin JDBC HDFS HBase S3 Swin 13

197. Zeppelin IntegraTon 14

198. Improved Performance and Stability •  Ooeap sort operator for ORDER BY (TAJO-‐907) •  Hash shuffle IO improvement (TAJO-‐374, TAJO-‐987) •  Skewness handling of hash shuffle •  AutomaYc parallel degree choice during runYme •  Lots of query opYmizer improvements •  Add Master HA (TAJO-‐704) •  More error-‐tolerant shuffle fetch (TAJO-‐789, TAJO-‐953) 15

199. What’s New in Tajo 0.11 16

200. Nested data and JSON support •  Nested data is becoming common •  JSON, BSON, XML, Protocol Buﬀer, Avro, Parquet, … •  Many web applicaYons in common use JSON. •  MongoDB by default uses JSON document •  Many Hbase users also store JSON document in a cell. •  Flakening causes lots of data/computaYon overhead. •  Tajo 0.11 naYvely supports nested data types. 17

201. How to create a nested schema table Use ‘RECORD’ keyword to deﬁne complex data type 18

202. Loose schema for self-‐describing formats You can handle schema evolving with ALTER ADD COLUMN! 19

203. How to retrieve nested ﬁelds Input Data Table DeﬁniYon SQL 20

204. Query federaTon and Tablespace support •  Query support across mulYple data sources •  You can perform join or union among tables on different systems. •  Benefits: •  Data offload from RDBMS to Hadoop vice versa •  A mixed use of exisYng RDBMS and Hadoop. •  Access to NoSQL and various storages through SQL •  An unified interface for SQL tools HDFS NoSQL S3 Swin Apache Tajo 21

205. Sequence File RCFile Protocol Buﬀer Data Formats Storage Types Datasets stored in Various Formats/Storages ORC 22

206. Tablespace •  Tablespace •  Registered storage space •  A table space is idenYfied by an unique URI •  ConfiguraYon and Policy shared in all tables in the same tablespace •  It allows users to reuse registered storages and their configuraYon. 23

207. Tablespace ConﬁguraTon Tablespace name Tablespace URI 24

208. Create Table on a speciﬁed Tablespace CREATE TABLE uptodate (key TEXT, …) TABLESPACE hbase1; CREATE TABLE archive (l_orderkey bigint, …) TABLESPACE warehouse USING text WITH (‘text.delimiter’ = ‘|’); Tablespace Name Format name 25

209. OperaTon Push Down SELECT X, SUM(Y) FROM table1 WHERE x 100 GROUP BY x Underlying Storage Filter, ProjecYon or Groupby can be pushed down into Underlying storages (like RDBMS, Hbase, ElasYcsearch, …) 26

210. Current Status of Storages •  Storages: •  HDFS support •  Amazon S3 and Openstack Swin •  Hbase Scanner and Writer -‐ HFile and Put Mode •  JDBC-‐based Scanner and Writer (Working) •  Kara Scanner (Patch Available) •  ElasYc Search (Patch Available) •  Data Formats •  Text, JSON, RCFile, SequenceFile, Avro, Parquet, and ORC (Patch Available) 27

211. Python UDF •  Python UDF and UDAF are supported in Tajo •  hkp://tajo.apache.org/docs/devel/funcYons/python.html @output_type('int4')  def return_one():  return 1    @output_type('text')  def helloworld():  return 'Hello, World’    @output_type('int4')  def sum_py(a,b):  return a+b 28

212. Get Involved! •  We are recruiYng contributors! •  General •  hkp://tajo.apache.org •  Gemng Started •  hkp://tajo.apache.org/docs/0.10.0/gemng_started.html •  Downloads •  hkp://tajo.apache.org/downloads.html •  Jira – Issue Tracker •  hkps://issues.apache.org/jira/browse/TAJO •  Join the mailing list •  dev-‐subscribe@tajo.apache.org •  issues-‐subscribe@tajo.apache.org 29

213. QA 30

Apache Tajo - An open source big data warehouse

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to Apache Tajo - An open source big data warehouse (20)

Recently uploaded (20)

Apache Tajo - An open source big data warehouse