Making Big Data a First Class citizen in the enterprise
Click here to load reader
-
Upload
tony-baer -
Category
Technology
-
view
612 -
download
1
description
Transcript of Making Big Data a First Class citizen in the enterprise
© Copyright Ovum. All rights reserved. Ovum is an Informa business.1
Making Big Data a First-class Citizen in the Enterprise
Tony Baer
IT014-002860
January 24, 2014
© Copyright Ovum. All rights reserved. Ovum is an Informa business.2
Contents
The premise
What we mean
Scope – focus on Hadoop, the most popular emerging Big Data platform
Addressing the enterprise
IT organization
Data center infrastructure, processes, policies, and practices
Creating value; competitive benefits to the business
The endgame
© Copyright Ovum. All rights reserved. Ovum is an Informa business.3
The premise: Voting Big Data off the island!
Big Data must become a first-class citizen in the enterprise and cannot exist on its own island.
© Copyright Ovum. All rights reserved. Ovum is an Informa business.4
Hadoop user base and use cases are changing
Users are changing from Internet companies to mainstream enterprises.
Use cases are changing from Internet search, ad optimization to customer churn analysis, sales and promotions, and operations.
© Copyright Ovum. All rights reserved. Ovum is an Informa business.5
What does it mean for Big Data to become a first-class citizen?
IT organization
No more SWAT teams! Must map to existing people and skills
Data center
Must map to existing infrastructure, subject to same constraints
Enterprise
Must address real business problems, not abstract data science research
© Copyright Ovum. All rights reserved. Ovum is an Informa business.6
Contents
The premise
Addressing the enterprise
IT organization
Data center infrastructure, processes, policies, and practices
Creating value; competitive benefits to the business
The endgame
© Copyright Ovum. All rights reserved. Ovum is an Informa business.7
Mapping Big Data skills to the IT organization
Enrich, don’t replace, your existing app developers, DBAs, system administrators
Huge existing SQL skills base – you’re not going to replace them
Large Java developer base, lots of scripting language diversity
Popularity of JavaScript/JSON
Skills:
Technology – the easy part
Domain and data science – not so easy
Don’t forget the people part!
© Copyright Ovum. All rights reserved. Ovum is an Informa business.8
Extending the IT organization for Big Data
SQL and NoSQL/Hadoop platforms are converging
SQL access to Hadoop
Hadoop platform SQL support
BI tool Hadoop support
MapReduce approaches to Advanced SQL platforms
MongoDB, CouchDB, Riak
Empowering web JavaScript developers with familiar JSON
Data science?
The apps are coming…
SQL on Hadoop and Big Data apps are works in progress…
© Copyright Ovum. All rights reserved. Ovum is an Informa business.9
Addressing the data center
Like most Internet technologies, Hadoop conceived in zone of trust
Small, elite band of practitioners
Big concern? Getting access to available cluster resources elsewhere inside the firewall
Enterprise?
Security
Data stewardship
Coping with finite resource
Availability and reliability
© Copyright Ovum. All rights reserved. Ovum is an Informa business.10
Data center: Big Data must be secure like any database management system or data warehouse
AAA enforced for access, authentication, authorization
Must become more granular by user, data
Must become more unified
Integrate with LDAP/Active Directory
Data privacy mandates
This is a policy, not a technology, issue
“Don’t be creepy” – don’t blindside your customers based on knowledge they didn't know you have
Regulation plays driving role for some sectors
© Copyright Ovum. All rights reserved. Ovum is an Informa business.11
Data center: Big Data platforms must behave like any database management system or data warehouse
Data stewardship/lifecycle
Data quality, protection, lifecycle management, retention
Resource management
Capacity utilization critical
Availability/reliability
Performance management essential for large clusters
Major change from early Internet adopters
© Copyright Ovum. All rights reserved. Ovum is an Informa business.12
Address the business
Good business cases count!
Do:
Focus on existing problems (the problems are often more obvious than you think…)
Identify key points of pain, like any new IT solution
Don’t
Concoct “interesting” data science problems for the heck of it
Get carried away with data (with lots of data, there are lots of chances for detecting irrelevant trends)
Give up after a few tries …. iterate!
Don’t get caught up in a data science project
© Copyright Ovum. All rights reserved. Ovum is an Informa business.13
Hadoop benefits: Solving familiar problems in new ways
Customer holistic view
Predictive churn analysis
upsell/cross-sell, next-best-offer,
cross-channel ID resolution
Risk mitigation
Fraud detection, counter-party risk
management, credit scoring
Operational efficiency
Machine data for managing smart
grids, smart urban infrastructure, supply chain
logistics
Not arbitrary data science
© Copyright Ovum. All rights reserved. Ovum is an Informa business.14
Contents
The premise
What we mean
Scope – focus on Hadoop, the most popular emerging Big Data platform
Addressing the enterprise
IT organization
Data center infrastructure, processes, policies, and practices
Creating value; competitive benefits to the business
The endgame
© Copyright Ovum. All rights reserved. Ovum is an Informa business.15
The endgame: What becoming a first-class citizen really means
Big Data – and emerging platforms like Hadoop – originated as specialized IT systems requiring specially skilled practitioners.
This model is not sustainable as Big Data crosses over to the enterprise.
Big Data must get off its island.
Big Data must be accessible to the IT organization, fit into the data center, and address real business problems.
© Copyright Ovum. All rights reserved. Ovum is an Informa business.16
Big Data: Embrace and extend
IT organization
Embrace existing SQL, Java, and other programming language skills
Extend skills to understand handling of larger volumes and varieties of data and new analytic techniques to supplement SQL
Data center
Embrace existing policies and practices for data stewardship, resource management, security, performance management
Extend policies and practices to accommodate platform with different workload characteristics, and support of active archiving
Business
Embrace existing competitive problems; don’t look for new problems because the data and platform are different
Extend approaches to problem solving by incorporating new data types and new forms of analyses to deepen understanding and insights