Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
1
May 2014
Big Data Summit
The Elephant in the Cloud
Changing Shape of Data
Confidential and Proprietary, Qubole Inc. page 4
Emergence of Hadoop
Confidential and Proprietary, Qubole Inc. page 4
Scalability on
Commodity Hardware
Democratization of
Data Processing
Impediments
Confidential and Proprietary, Qubole Inc. page 4
Investment Risk as upfront investment needed to discover value of data
Execution Risk in order to come up to speed on the technology and then integrating it
Operationalizing this new technology as use case move into production
Enabling Accessibility to this technology in the enterprise
Benefits of the Cloud
Confidential and Proprietary, Qubole Inc. page 4
On-demand and Turn Key without the hassles
Flexible in supporting different workloads and use cases and growing as
the enterprise moves from PoC to Production
Accessible in multiple regions and geographies
6
Use Hadoop on the Cloud to Discover and
Accelerate Big Data Use Cases
Cloud and Big Data
Confidential and Proprietary, Qubole Inc. page 6
De-risk Hadoop & Big Data
Confidential and Proprietary, Qubole Inc. page 5
80 node PoC Cluster to start with
that grew to 3000 node Cluster
Platform from 2007 to 2011
Took 3 months to get the cluster in
place and the base software
deployed on it
Took another 9 months to make it a
true strategic platform
Cloud based Hadoop as Service
Confidential and Proprietary, Qubole Inc. page 5
Zero upfront investment in
infrastructure
Instantaneous scaling because of
the cloud
Hadoop becomes a strategic
platform very quickly
De-risk Hadoop & Big Data
Confidential and Proprietary, Qubole Inc. page 5
System Mgmt
Hadoop
Scheduler
Hive/PIG
MonitoringGUI
Interfaces
(ODBC/JDBC)
Data Connectors
Lots of moving parts and open source technologies needed before getting to a data processing platform
Cloud based Hadoop as Service
Confidential and Proprietary, Qubole Inc. page 5
System Mgmt Hadoop Scheduler(Oozie)Hive/PIG
Mahout/Weka
MonitoringGUI(Hue) Interfaces
(ODBC/JDBC)
Data Connectors
(MongoAdaptor..)
Fully Integrated Turn Key Platform that enables you to discover the ROI quickly with low execution
and investment risk
Challenges in Operationalizing Hadoop
Confidential and Proprietary, Qubole Inc. page 5
Managing Growth placed on the infrastructure as usage grows
Managing Unpredictability in today’s agile development environment where use cases
change leading to different types of demands from the infrastructure
Managing Open Source as there is constant innovation and internal operations teams
have to continue to monitor open source contributions and distros
Cloud based Hadoop as a Service
Confidential and Proprietary, Qubole Inc. page 5
Self Managed the growth of the infrastructure is matched with the growth of usage
and data dynamically and in an on demand manner
Flexibility and Elasticity of the cloud ensures instant availability of different machine
types for different use cases
Open Source Innovation is managed by the service provider while enterprises can focus
on deriving actionable insights from their data
Cloud Accessibility
Confidential and Proprietary, Qubole Inc. page 5
Cloud based Hadoop as a Service
Confidential and Proprietary, Qubole Inc. page 5
Sharing and Collaboration of data and analysis
is easily enabled across:
Geographies
Organizations
Supply Chains
Objections to a Cloud Service
Confidential and Proprietary, Qubole Inc. page 5
Is TCO higher in the rent model vs pay upfront model?
Prices on the cloud keep dropping due to economies of scale and competition
Storage: 3 cents/GB/month (70% drop in the last 4 weeks)
Accenture study found broadly better performance in the same price on the cloud
http://www.accenture.com/SiteCollectionDocuments/PDF/Accenture-
Hadoop-Deployment-Comparison-Study.pdf
It is already more cost effective to run on the cloud and it is just going to get better
Objections to a Cloud Service
Confidential and Proprietary, Qubole Inc. page 5
Is security and compliance there?
For security use encryption
For compliance AWS has many new products such as auditability,
Even the CIA runs workloads on the cloud
http://www.crn.com/news/cloud/240163382/amazon-wins-600-million-cia-
cloud-deal-as-ibm-withdraws-protest.htm
17
Cloud Based Hadoop Services provide a
Quick path to a Big Data Platform that
Adapts to the needs of an organization while
Reducing Costs and
Reducing Failure Risk
Cloud and Big Data
Confidential and Proprietary, Qubole Inc. page 6