1. Nov 2006 Google released the paper on BigTable.
Feb 2007 Initial HBase prototype was created as a Hadoop contribution.
Oct 2007 The first usable HBase along with Hadoop was released.
Jan 2008 HBase became the sub project of Hadoop.
Oct 2008 HBase was released.
Jan 2009 HBase was released.
Sept 2009 HBase was released.
May 2010 HBase became Apache top-level project.
2011 Hbase 0.92

2. Cloudera Image for hands-on
Installation instruction

3. Agenda
Now
HBase architecture
Data operations - hands on
Summary

4. ? Recap Mahout Sqoop Oozie Hive Pig Spark Impala Flume Hbase Zookeeper
Sequential data scanning with Scala, Python, Java, SQL
Sequential data scanning with Java
Sequential data scanning with SQL
using MapReduce
Sequential data scanning with SQL (direct data access)
Machine learning
Mahout
Workflow manager
Oozie
Data exchange with RDBMS
Sqoop
Scripting
Pig
Hive
SQL
Large scale data proceesing
Spark
Distributed file system
Log data collector
Flume
Impala
SQL
NoSql columnar store
Hbase
Zookeeper
Coordination
MapReduce
YARN
Cluster resource manager ?
HDFS
Hadoop Distributed File System

5. What is HBase? NoSQL database on Hadoop Optimized for random reads
Key – value store, schema-less
For storing big tables with many rows and columns
Consistent inserts, updates and deletes of rows
Optimized for random reads
Data partitioning by row key values
Index on row key values
Bloom filter
Column store
Scalable

6. What HBase is not? Not a relational database Transactions are not ACID
Index available only on a row key
Weak for sequential data scanning

7. When to use?
In general:
For data too big to store on some central storage
For random data access: quick lookups of individual records
The data can be represented by key-value sets
Database of binary records (serialized objects, documents)
When data set
has to be updated
is sparse – records have variable number of attributes
has custom data types (serialization)

8. When NOT to use? For massive data processing/analytics
use MR, Spark, Hive, Impala… instead
For data sets with very high frequency insertion rates
stability concerns - from own experience
Data schema is complex
If “I do not know what solution to use”