A Methodological Approach to Big Data Proof-Of-Concepts ... Methodological Approach to... ·...

Post on 22-Mar-2018

214 views 2 download

transcript

A Methodological

Approach to Big Data

Proof-Of-Concepts for

Business & IT

Boubkeur KAAOIASS

Business Perception

IT Perception

Business & IT

challenges

How to approach the

Elephant ?

What are the pitfalls?

Building a sustainable

master plan/roadmap

Topics

Business – Sees Opportunities

For Business a complete new promising

& exciting horizon of possibilities

IT – Sees Problems

For IT a frightening journey to an unknown land ruled by a giant elephant

Hadoop Extended Ecosystem

Impressive Technology Stack

Complementing & Overlapping in

functionalities

Appearing/Disappearing/Changing virtually

everyday!

Main Business Challenges

Agree on Use Cases

Identify Business Stakeholders

Locate Data

Evaluate Relevancy

Prove Value

Main IT Challenges

Agree on Technical Use Cases

Identify IT Stakeholders

Technology & Platform Selection

Supplier Selection

Build in-house competencies

How to glue the Business initiative with the IT initiative?

• How to make sure the solution that IT will recommend will be able to support the business use cases?

• What to POC on the IT side ?, all technologies not feasible

• A subset ? risk of not covering business requirements

• How to make sure IT will be able to feed the value chain in terms of data?

• Business POC will mimic the data & the process what if production data are not available in due format & time.

• How to align planning?

• Usually initiatives don’t have the same timelines & milestones.

• Planned independently

• How to align big data initiatives?

• We see often business discussing directly with vendors without their IT involved/informed

4 Simple questions but which could have a huge impact if not tackled properly and in due time; right from the very beginning.

Ultimately determine the success or the failure of your big data initiatives

But…..

•Know your Enterprise Maturity

• Share a common platform

• Appoint a “man-in-the-middle”

• Apply Agile Methodology

• Focus on Business value

• Have your IT up-to-speed

• Leverage the Cloud

• Share a common roadmap

Approach To Secure Success

Know your Maturity!

Capitalize on your

Competence

Center

Discuss, Agree &

Decide on key

aspects

Monitor & Report

Regularly

Escalate fast

Embark Data &

Security Officers

Share a common platform

The guy who makes sure that all

stakeholders are ALIGNED!

Focus on Communication

Man-In-The-Middle

Go Agile: Scrum it!

Assign Big Data Agility Roles &

Responsibilities

Don’t put the technology in the driving seat

Don’t forget your Enterprise technology base

Think Big but Start Small

Apply Fail-Fast Principle

Value is not in what others do

Business Value is everything

Streaming is cool but do you really need it?

Does your use case really need Hadoop?

Don’t try to “big bang” your organization

Real-Time Offering is cool but will you increase your customer portfolio?

Have your IT up-to-speed

Involve your IT teams not in supporting role of vendors and suppliers but let them do the job

Foresee training & coaching from the very beginning

Think Hadoop Lab

Don’t put to much pressure

Leverage the Cloud

Because the average Duration of the POC: 3-4 months

It’s impossible to install, setup, configure and test all Big Data solutions:

Hadoop & Distributions (e.g. Hortonworks & MapR) Big Data Solutions (e.g. Teradata Aster & IBM BigInsights)

Think on the possibility to deploy them on the cloud

Use available benchmarks to speed the evaluations (.e.g. HiBench)

Micro Benchmarks (Sort, WordCount, TeraSort,…)

Web Search (Nutch Indexing, Page Ranking)

Machine Learning (Bayesian Classification, K-Means Clustering)

HDFS Benchmarks (I/O)

Share a common roadmap

Next

2016

2015

2015

2015

Based on POCs experience and previous list of eligible big data projects, define a detailed road map. Criteria of choice : emergency, Expected ROI, Team availability ….

Define and prioritize the projects eligible to big data project Run one or two project in a POC mode to adjust/validate the architecture, the organization and governance

Include the Big Data governance in the global data governance program as defined previously Define the roles Appoint people Change management organization

Move existing platform to hadoop environment ( e.g : Info Centers) based on Enterprise hadoop experience and Market maturity

With the experience of the IT POC and taking account the business pilots , choose the best platform for the Enterprise Define development rules

Big data centric IS

Industrialization

Governance definition

Business POC

Hadoop Platform choice

Work Hard but Have Fun!

Boubkeur KAAOIASS:

boubkeur.kaaoiass@arhs-data.com

Senior BI consultant

Gunther ROOBAERT:

gunther.roobaert@arhs-data.com

+32 475 67 30 21

Director Arhs Data

Join us at the booth today

Talk to us