data architectplan.medone.co.kr/38_ghrforum/data/c-2chang.pdf · 2019-06-26 · - data architect is...
TRANSCRIPT
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Data Architect : Roles, Requirements and Values as Future Job
Sungwoo Chang Lead, Solution Engineering Oracle Korea
Global HR Forum 2018
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Agenda
Change in Data Management
Data Architect as Future Prospect
Roles and Requirements
Suggestions
2
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Change in Data Management
3
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. | 4
Needs for Data Management and Analytics Everywhere
Product Requirements/ Specifications Mgmt
Opportunity Status
Delivery Order
6
Logistics Mgmt
Product Quality, Customer Claims & Warranty Mgmt
Planning
Product Lifecycle Mgmt
Marketing & Sales Mgmt
Pre-Sales and Sales Activities
Supply Chain & Production Planning
Production Mgmnt & Manufacturing
Transportation Plan and Execution
Delivery
After-sales Services
Finance & Accounting
Human Resources
Management
Yield Rate Improvement
Early detection of Product Defects
Product Usage Monitoring & Advices
Procurement Cost Reduction
VoC Analysis
Marketing & Sales Activities Optimization
Work-Style Reform
Demand/Sales Forecast
Quality Mgmt & Assurance
Maintenance Optimization & Equipment Failure Prevention
Product Failure Predicts with Proactive Service Actions
Productivity Improvement
Employee Satisfaction Improvement
Global Cost Mgmt
Best offer analysis
Social Listening Detailed Mgmt
Simulation
Decision Making Support
Increasing needs of decision making based on data everywhere Voice of Customer Voice of Product
Voice of Factory
Voice of Enterprise
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Data Architecture Considerations RDB Only RDB + NoSQL + Hadoop + α : “ Best Together ”
Relational Database Hadoop
Best Use for :
Business Data (Account, Customer etc)
Rigid transaction processing(ACID)
Guaranteeing logical consistency and physical safety for multiple users concurrent access
100% SQL Compliance
High cost
Best Use for :
Sparse data such as web/sensor logs
Parallel Batch Processing
no support for transaction processing
Fit for data collection and preprocessing
Low cost
NoSQL
Best Use for :
Data for SNS, Blog etc
Partial Consistency(Delay allowed)
Flexible and efficient
Designed for special purpose
Complimentary with RDBMS
Tips for TCO Reduction : Storing Data at Appropriate Systems based on the Characteristics of Data
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Future Data Architecture
`
Traditional Data
Customer Product
Revenue Order
Contract
Call Center ….
Emerging Data
SNS
Web
Mobile
Sensor
Log
Ext. Data
….
Realtim
e
Layer
Streaming
Ingestion
Real-Time
Machine learning Model
Real-Time Monitoring
Scala
Big Data Lake
Hadoop Hive Hbase Spark Impala Oozie Zookeeper
Solr
Hue
Landing Staging
내부
외부
기준 정보
분석/연계 모델
분석 레파지토리
Analytical Sandbox
Big Data SQL
Agile Mart Service Mart
AI (CDSW)
Machine/Deep Learning
GPU CPU
Batch
Layer
AI - Machine & Deep Learning based Data Platform
AI-enabled
• Adopting ML/DL on big data for prediction purpose
40G (전처리 결과)
Big Data Lake
Analytical Sandbox
Common Service Layer
• “Source Data Store “
• Landing Area to process Hadoop-based data suitable to analytical purpose
• “Free Analytical Space“
• For Analyst using Big Data SQL to access Hadoop and RDB in an integrated way
• “Space for getting business-oriented analytical information
• Statistical analysis is required
• Visualization is critical
• A number of highly trained people are required to design, implement and manage the data architecture
Copyright © 2018, Oracle and/or its affiliates. All rights reserved. 7 Copyright © 2018, Oracle and/or its affiliates. All rights reserved.
Self-Driving
Self-Repairing
Self-Securing
Reduces human labor
Protects itself from attacks
Keeps business up & running
Data management also goes to ‘Autonomous’
New Trends : Autonomous Database
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Data Architect
: Roles and Requirements
8
Copyright © 2018, Oracle and/or its affiliates. All rights reserved. 9 Copyright © 2018, Oracle and/or its affiliates. All rights reserved.
Changing Roles in Data Management
Free from Simple and Repetitive Jobs More Requirement for Creative Work
Reducing Maintenance Work Increasing Innovative Work
• Less time for infra management
• Less effort for patch and upgrade
• Less work for availability design
• Less work for system tuning
• More time for data architecture design
• More work for new application
development in very short time
• More labor for data analytics
• More investment to enhance data
collection and quality evaluation
Copyright © 2018, Oracle and/or its affiliates. All rights reserved. 10
From DBA to Data Architect To-Be
“Data Architect”
Focusing on Business Support as Data Owner
• Designing optimized data
architecture including cloud
technology
• Defining security and ILM process
• Business support based on data
analytics
• Close work with developers
As-Is “DBA”
Focusing on
DB Maintenance
• Install, Operation,
Monitoring
• Backup/Recovery and
Trouble-shooting
• Relatively low portion of
business-related work
Transition
Trainings for
Transition
• EDA(Enterprise Data
Architecture) training
• Big data technologies -
Hadoop, Spark, NoSQL
• Demo & Hands-On capabilities
for various cloud services
• Emerging technologies such as
AI, Block-chain, IoT etc
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Requirements • Spectrum of technologies required to be a Data Architect
Copyright © 2018, Oracle and/or its affiliates. All rights reserved. | Copyright © 2018, Oracle and/or its affiliates. All rights reserved. |
Suggestions
Balanced Renovations in 4 Key Pillars
Technology, Architecture, Competency, Leadership
Sensemaking
Data Analytics with Humanities
12
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Four Areas where Renovations are Required - Technology, System, Personal Competency and Leadership
• Technology
– All-Data Architecture : RDB + NoSQL + Hadoop
Best Together
– Spark with AI for real-time analytics
• Architecture
– Micro Services Architecture with Cloud Adoption
• Personal Competencies
– Data Architect with Sensemaking
• Leadership
– Productive team coming up with good services based data analytics and customer understanding
– New Leadership : Google ‘Project Oxygen’
Copyright © 2017, Oracle and/or its affiliates. All rights reserved. |
Book Review : Sensemaking
• What is Sensemaking? – A way of getting practical knowledge based on Humanities
– Perceiving objects not just from themselves but from relationships of themselves
Understanding context of data is required as well as data themselves
• Five Ways of Sensemaking – Culture, not Individuals
– Thick data, not just thin data
– The Savannah, not the Zoo
– Creativity, not Manufacturing
– The North Star, not the GPS
Copyright © 2018, Oracle and/or its affiliates. All rights reserved. |
Summary - Data Architect is the Foundation of all in the age of Data -
Change in Data Management
New Data Architecture : RDB + NoSQL + Hadoop with On-Premise/Cloud mixture
Changing Roles in Data Management
From DBA with micro and system-focused jobs
To Data Architect with macro and business-oriented creative works
Suggestion
Not only personal competencies but also team work and new leadership