2. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Apache Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
3. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
4. Batch to Real Time
Changing image of Big Data from Batch to Real Time
Hadoop + MapReduce = Batch Processing
5. Batch to Real Time
● Companies need real time processing of Big Data for
various applications including online Fraud Detection,
CEP (Complex Event Processing) and more.
● Emerging new frameworks, architectures and tools are
making the real time processing dream come true.
6. Big Data Real-Time Computing Systems
● Twitter’s Storm is an open source, distributed, faulttolerant and real time computation system.
○ Storm is a stream processing system
○ Unlike Hadoop jobs Strom jobs never stop continue
to process data as it arrives
● Other Real Time systems include Streambase,
HStreaming, Apache S4, Dempsy and Esper.
7. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
8. Big Data Sql Tools
Big Data Processing include ...
● Writing complex Java MapReduce Jobs
● Apache Pig Latin scripting
● Slow Sql processing from Apache Hive
9. Big Data Sql Tools
Inspired with Google’s Dremel paper now many vendors
offer faster SQL based tools
● Google BigQuery
● Cloudera Impala
● IBM BigSql
● Greenplum HAWQ
● Hortonworks Stinger (Improve Hive Sql by x100)
● Apache Drill
10. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
11. Big Data And Cloud
Big Data needs many computing nodes for Data Storage
and Data Processing which are elastic in nature …
● Cloud VM based computing is a perfect solution for
Big Data infrastructure
● Public Cloud MegaStar Amazon AWS announced
support for Hadoop, which means spin off Hadoop
installed VM with basic configuration in 10mins
12. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
13. Hadoop 2.0
New in Hadoop 2x
● Improved Performance with YARN aka MapReduce 2.0
● Improved Scalability with HDFS Federation
● Support for Microsoft Windows
● Improved Security
● HDFS Snapshots
14. Hadoop 2.0 - Performance
Improved Performance with YARN aka MapReduce 2.0
● MapReduce JobTracker managed both Resource
management and App Job life-cycle together before.
● Now two functions are divided into separate
components.
● Application Master negotiates with global Resource
Manager for various Job requests
15. Hadoop 2.0 - Scalability
HDFS Federation
● No more single NameNode(NN) and SNN.
● HDFS Federation supports multiple independent
NameNodes and Namespaces.
● Each DataNode(DN) registers with all the NameNodes in
the cluster. DN sends periodic heartbeats & block
reports and handle commands from all NN.
16. Hadoop 2.0 - Security
Improved Security
● Enforcement of HDFS file permission by NN and Access
Control List (ACL) of users and groups
● Block Access Tokens for access control to Data block.
● Job Tokens to enforce Task authorization
● Network Encryption & Kerberos RPC. Now HDFS file
transfer can be configured for encryption
17. Hadoop 2.0 - HDFS Snapshots
Improved Backup & Disaster Recovery
● HDFS Snapshots are read-only point-in-time copies of
the file system.
● Snapshots can be taken on a subtree or entire file
system.
● Useful for data backup, protection against user errors
and disaster recovery
18. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
19. Big Data Applications
● Infrastructure layer of Big Data is largely solved (.........
secret Hadoop)
● Now the future innovation is focused on applications and
analytics
20. Big Data Analytic Applications
Pattern Discovery and Sense-Making based analytic
applications.
● Wibi Data: Lessons learned and predictive apps
● Recorded Future: Web intelligence for Business decisions
● Nutonian: Uncovers relationships hidden with in complex
data
● R Studio: Data analysis tool
21. Big Data - Visualization Applications
Sophisticated Big Data Visualization tools.
● IBM BigSheets
● D3.js
● Fathom
● Processing.org
22. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
23. Big Data & Business Intelligence
Support from various BI vendors IBM Cognos, SAP Business
Objects & Oracle Hyperion to connect directly to Hadoop Data
using Apache Hive connectors.
24. Big Data & Data Warehouse
Challenge of new multiple unstructured data sources such as
Clickstreams, Social media, Mobile, Sensors and Web Logs
requires massive processing and traditional data warehouse
cost to scale.
The Big question is data warehouse survive the Big Data ?
More on this in my next presentation :)
25. Agenda - Big Data Trends
●
●
●
●
Batch to Real Time
Sql, Sql, Sql …
Cloud Platform Support
Hadoop 2.0
○
○
○
Improved Performance
Improved Scalability
Improved Security
● Applications
○
○
Pattern Discovery Analytics
Sophisticated Visualization
● BI & Data Warehouse
● Big Data Vision
27. Big Data requires Big Vision
● Unlike Business Intelligence, Big Data is an innovation
originated from the IT side.
● The Business departments, which should come up with Big
Data usage requirements needs constant coaching on the
potential of the Big Data intelligence and successful
stories.