View
5
Download
0
Category
Preview:
Citation preview
IBM Cloud Forum
20 novembre 2019New Cap Event Center, Paris
Exploiter toute la valeur de vos données
avec Cloud Pak for Data
Cloud Pak for Data :La plateforme collaborative de services Data & AI
2 11/20/2019
Corinne Baragoin
Data Architect
c_baragoin@fr.ibm.com
IBM
3 11/20/2019
82% are concerned with [data] connectivity across cloudsDataAI
Cloud
66% of cloud
workloads will be AI-driven
94% use multiple clouds platforms
Data is what fuels digital transformation
Digital transformation requires unified Data + AI + Cloud services spanning an open, hybrid cloud environment
Source: IBM MD&I; : BCG and McKinsey, MIT Slone, Forrester, LogicMonitor
Modern Architecture
4
A faster, more secure way to move your core business applications to any cloudthrough enterprise-ready containerized software solutions
Cloud Paks – Enterprise-ready containerized software
IBM containerized softwarePackaged with Open Source components,
pre-integrated with the common operational services,and secure by design
Container platformand operational services
Logging, monitoring, security,identity access management
IBM Cloud Private SystemsEdge
Complete yet simpleApplication, data and AI services that aremodular, term licensed, and easy to consume
IBM certifiedFull software stack support, and ongoing security, compliance and version compatibility
Run anywhereOn-premises, on private and public clouds,and in pre-integrated systems
5
Cloud Paks – Pre-integrated for cloud use cases
Today, IBM offers clients the first six Cloud Paks…
IBM Cloud SystemsEdge Private
Cloud Pak for Security
Containerplatform andoperational services
Cloud Pak for Applications
Developer & DevOps Tools
ModernizationToolkit
Frameworks and Runtimes
Containerplatform andoperational services
Cloud Pak for Data
Containerplatform andoperational services
Organize Analyze
Collect
Cloud Pak for Integration
Containerplatform andoperational services
API Lifecycle
Messaging and Events
App and Data Integration
Cloud Pak for Automation
Containerplatform andoperational services
Workflow and Decisions
Operational Intelligence
Content
Cloud Pak for Multicloud
Management
Containerplatform andoperational services
App and Infrastructure
Multicluster
Security and ComplianceManagement
Incident Response
Federated Search and
Investigation
Security Orchestration and Automation
6
Data Ecosystem• Data in silos• Difficult to access• No lineage
Analytics Tools• Discrete tools
• Different preferences
• Difficult to manage
Workflow• Not integrated
• Not governed
• Lack dev/prod parity
Culture• Not collaborative
• Slow provisioning
• Lack trust in AI
“No amount of AI algorithmic sophistication will overcome a
lack of data [architecture] … bad data is simply paralyzing”
Data & AI Challenges
Modernize your applications
7
AppDeployment
Data
New Requirements& Engagement
DevelopmentTest
MonitoringRetraining
Search for Data
Acquiring Data/Self Service
Modelbuilding
Hadoop
EDWNOSQL
Data Lake
ML Model Deployment
RefiningData
Continuous Delivery
of Applications
Continuous Delivery
of Insights
Multi-CloudGovernanceMicroservices & APIs
7
COLLECT - Make data simple and accessible
ORGANIZE - Create a business-ready analytics foundation
ANALYZE - Build and scale AI with trust & transparency
Data of every type, regardless of where it lives
INFUSE - Operationalize AI throughout the business
AI
MODERNIZEUnlock the value of data for an AI and multicloud world
One Platform, Any Cloud
The IBM Data & AI Ladder
8
9
The IBM Data & AI Offering
Watson Data Science Platform (DS/AI)Watson Studio, Watson Knowledge Catalog, Watson ML, Watson OpenScale, Decision Optimization, SPSS,
Watson Apps & SolutionsWatson Assistant, Watson Discovery, Watson APIs, Cognos Analytics, Planning Analytics, BPM, GBS
Integration & GovernanceInfosphere Family, Watson Knowledge Catalog
Hybrid Data ManagementDb2 Family, Hadoop/Cloudera, MongoDB
LOB Execs and Business Analysts
Data Stewards, Data Engineers,
Compliance Officers
Data Engineers, Data Developers
and Data Admins
Data Scientists, AI Developers,
Programmatic Analysts
Cloud Pak for DataCIO, CTO, CDO, Cloud
& Data Architects
Speed time to value with prebuilt AI apps
Collect and manage hybrid Data of all types
Deploy a unified Data & AI Cloud Platform
One Platform, Any Cloud
Infuse
IBM Cloud Pak for Data : Collaborate
App DeveloperData ScientistData Engineer
Enterprise Catalog
✓ Integrate & Refine
✓ Deploy models as scalable web services
Data Steward
Web service
Applications Business processes
Business Analyst
AnalyzeOrganizeCollect & Connect
Watson Services
Make your data ready for an AI and hybrid multi-cloud world
Make data simple andaccessible
Create a business-ready analytics foundation
Build and scale AI with trust and explainability
Operationalize AIthroughout the business
✓ Eliminate data silos
✓ Connect all data
✓ Utilize fit for workloaddata repositories
Virtualize
✓ Automate and governthe data & AI lifecycle
✓ Analyze using open sourceand visual tooling
✓ Explore and visualize
✓ Build analytical models
✓ Operationalize AI
✓ Measure and track AI outcomes✓ Index and enrich assets
Multicloud Services• Logging
• Monitoring
• Metering
• Persistent Storage
• Identity Access Mgmt.
• Docker Registry / Helm
• Kubernetes
• Security
© 2019 IBM Corporation
IBM Cloud Pak for Data
Unified platform of foundational Data & AI cloud services
IBM IBM BPIBM OSISVDSP Custom
On-Premise
KUBERNETES BASEDContainerized, easy to manage
PICK YOUR CLOUDPrivate or Public
PICK YOUR ADD-ONContainerized Services
DATA PLATFORM#1 Ranked by Forrester
Customize & Extend with add-on microservices
Built for Multi-cloudAvoid vendor lock-in & get started on your cloud journey today
11
IBM Cloud Pak for Data Key Principles
• Omnipresent, yet invisible – infused throughoutGovernance
• Common look-and-feel with customizable persona-based workflows
• Automated common services – user management, authentication models, security configurations, provisioning, collaboration, etc.
Pre-Integrated Experiences
• Cloud-native architecture
• Cloud agnostic & multi-cloud – any vendor cloud or data center
• Hyper converged System available
• Coherent, efficient, and scalable data & analytics services
Deploy Anywhere
12
Open by Design
A modern information architecture meets open source, on any cloud, working as one
The innovation & skills of Open Source Communities
IBM Cloud
One Open Platform, Any CloudBuilt upon cloud, data and AI open source frameworks
15 11/20/2019
Data virtualization
Db2 Warehouse Governance
Data Discovery
Data Visualization & Dashboards Data Science: Model Deployment
1. Cloud Pak for Data ServicesCollect Organize Deploy
Powered by: new Db2 Technology & Db2 Warehouse
Powered by: Information Analyzer, IGC & Data Stage Powered by: Watson Studio
Db2 AESEDb2 BigSQL
Infosphere DataStage for CPDInfosphere Regulatory AcceleratorInfosphere multi-cloud Data MvmntInfosphere Entity Resolution
2. Premium Cartridge Services (Purchase license or BYOL)Collect Organize Infuse
Analyze Powered by: Cognos CDE
Cognos AnalyticsWatson Studio Premium
(SPSS Modeler, Model builder, Decision Optimization, Watson Explorer, Streams Designer)
Analyze
Auto AIWatson Bundles
(Discovery, Assistant, Speech to Text, Natural Language Understanding, API Kit,Watson Knowledge Studio)
3. Third Party Add-Ons
IBM Streams Db2 Event Store
PostgreSQLData Science : Model Monitor with
Open Scale
Watson Knowledge Catalog
• Spark• Python 2.7 with Anaconda, R, Scala• Python 3.5 with Tensorflow GPU• Apache Zeppelin• Rstudio etc.
Data
Science
Model Build
Foundation• Logging• Monitoring
• Metering• Persistent volume /Storage
• Identity Access Mgmt.• Docker registry/Helm chart mgmt.
• Kubernetes• Security
Cloud Pak for Data Packaging
16 11/20/2019
Use Cases
Operationalize Data Science & AI
e.g. accelerate GDPR Compliance
Build, deploy, manage & govern models at scale to improve business outcomes
e.g. a. Customer Churn
b. Cross Sell / Up Sell
c. Predictive Maintenance
Shift to Cloud Native
a. Provision & scale in minutes
b. Build once, deploy anywhere – multi cloud support
c. Built in automation & collaboration to increase productivity
1. Manage all your enterprise data regardless of where it lives
(Data Virtualization)
2. Gain control & leverage your data from connected devices
(Fast data & Streaming analytics)
Manage your Data Anywhere
Shift to Next-Gen Workloads Smarter Governance
Governance to enable self service analytics
Auto-discover meta data, manage governance rules & policies, enforce privacy etc. to mitigate risk & ensure compliance
Data
Architects
ETL
Developers
Data Engineering Data Governance Teams Data Consumers (LOB)
DG Technical Users DG Business Users Data Scientists Business Users
Quality
Developers
Repository
Managers
Quality
Developers
Data
Stewards
LOB
RiskCDO
LOB
Product
Data
Scientists
Data
Analysts
Business
Analysts
Data Architects and Engineers
IBM Cloud Pak for Data
Data Ingestion Data Transformation Data QualityData Governance
Technical UsersData Governance
Business UsersActivate and Exploit The Data
• Search and find relevant data
• Data Preparation
• Consume and analyze the data
• Comment, rate and share
• Business lineage*
• Reference data management*
• Data ownership
• Data stewardship
• Data governance workflow
• Profile data
• Understand data quality
• Classify data
• Build validation rules
• Apply validation rules
• Monitor data quality
• Remediate data quality
• Extract data
• Collect metadata
• Move data
• Ingest data
• Build integration jobs
• Run integration jobs
• Monitor
• Discover metadata assets
• Classify data assets
• Build data glossary
• Create data lineage
• Manage metadata repository
Get Business Ready Data
Understand Your Data AnywhereWatson Knowledge CatalogOne Application. Delivering trusted data to the Enterprise.
WatsonKnowledge CatalogQuality | Governance | Catalog
Data LineageOperational Lineage
Asset LineageAI Lineage
Self ServiceShopping for dataData Preparation
Advanced Data Discovery
Data Source DetectionBusiness Term DetectionAdvanced Classification
Data QualityData Rules
Data Quality AnalysisDashboards
Business GlossaryTerm Management
Hierarchy managementReference Data
GovernancePolicy ManagementGovernance Rules
Policy Enforcement
© 2019 IBM Corporation 18
Enterprise Data
Integration
Enterprise Data
Governance
Enterprise Data
Quality
Enterprise Data
Consumption
Business Ready Data Foundation
19
The Ladder to AI
Multicloud Data & AI Platform
IBM Cloud Pak for Data
Unlock the value of your data and accelerate your journey to AI
Everything you need for enterprise data science and AI
AutoAI Lifecycle Automation – “AI developing AI”
Watson
Studio
Watson
MachineLearning
Watson
OpenScale
Watson
Knowledge
Catalog
Data Profiling & Prep
Quality & Lineage
Policy-based Governance
Visual Design
Develop & Train
Lifecycle Mgmt
Run & Optimize
Model-ops
Dynamic Retraining
KPIs & Accuracy
Explainability & Lineage
Automated Optimization
Prepare and Organize Data
Build and TrainAI Models
Deploy and Run AI Models
Manage and Operate Trusted AI
Automate and Industrialize AI
ICP4Data System
Use Cases
Watson Services & Applications
Pre-Integrated Tools, Algorithms, Librariesfor Data Science, ML/DL
• Best-of-breed tooling from open ecosystem• Authoring tools• Machine learning, deep learning, optimization• Customize environments, packages & images
• Coding and visual modeling options
• Cloud Pak infrastructure• Container-based resource management• Scale with distributed and GPU support
• Model Lifecycle Management• Dev -> Test -> Staging -> Prod• Versioning, release, SLAs, rolling upgrades
Governed AI Lifecycle Management on Cloud Pak for Data
Data Exploration
Data Preparation
Model Development
Build
Watson Studio
Watson Machine Learning
Business KPIs and production
metrics
Fairness & Explainability
Inputs for Continuous Evolution
Infuse
Run
Deployment
RetrainingModel
Management
Watson OpenScale
Rebuild models, improve
performance and mitigate bias
Monitor and
orchestrate
models served
with WML
Easily deploy
models to WML for
online, batch,
streaming
deployments
What does AutoAI do?
• Integrated with Watson Studio and Watson Machine learning
• Automatically ingest, clean, transform, and model with hyperparameter optimization
• Training feedback visualizations provide real-time results to see model performance
• One-click deployment to Watson Machine Learning
https://www.ibm.com/demos/collection/IBM-Watson-Studio-AutoAI/
AutoAI : automation of machine learning tasks
Watson APIs Suite on Cloud Pak for Data
Cloud Pak for Data
Wat
son
Ass
ista
nt
Wat
son
Dis
cove
ry
Wat
son
AP
I Kit
Kn
ow
led
ge S
tud
io
Speech To Text
Natural Language Understanding
Watson Knowledge Studio
Text To Speech
24 Cloud Analyst Summit/ Data & AI/ July 2019 / © 2019 IBM Corporation
USE CASE
Provide Nedbank’s customers access to fully functional ATMs at all times
• Better predict cash-outs & machine failures
• Optimize service and cash replenishment schedules
Optimize Nedbank’s ATM Experience
CASE STUDY“Even with a team of experienced data scientists on the ground, IBM was able to augment my team, provide strong technical leadership, and put in place a strong practice to set us up for quick delivery, but also enabling us for success in the future.”
Guy Taylor, Head of Data & Data-Driven Intelligence,
Nedbank
EXPECTED BENEFIT
Improve customer experience
Reduce planning cycle
Reduce replenishment & service costs
UNIQUE CHALLENGE
Difficult to predict machine fault category
Lengthy planning cycle due to uncertain travel times, custodian skills & availability
24
25 Cloud Analyst Summit/ Data & AI/ July 2019 / © 2019 IBM Corporation
Turbo Charged Digital Transformation
Sprint chose Cloud Pak for Data because it enables AI projects in weeks rather than months through unifying and simplifying three critical stages in the journey to AI: the collection, organization and analysis of data.
Facing its own path toward digital transformation, Sprint started preparing its data for Artificial Intelligence (AI) – with the goal of using machine learning algorithms to gain quicker insights and increase responsiveness to customers.
Data and AI / © 2019 IBM Corporation
“ Include a quote if one is available, we can always add this at closeout. ”
-- Include quote author’s name and title
“ Include a quote if one is available, we can always add this at closeout. ”
-- Include quote author’s name and title
Solution
Cloud Pak for Data
Read the story in any of these magazines: Business Chief US , Business Chief Canada , Gigabit Magazine
Modernize
Unlock the value of your data for an AI and multicloud world
Michelle Gehl
VP Networks OSS
Applications and Operations,
Sprint
IBM Cloud Pak for Data enabled Sprint to digest high volumes of data for near, real-time ML/AI analysis, and the trial results have shown potential to take Sprint to the next phase of digital transformation.
Industry: TelecommunicationsGeography: North America
Sprint
26 Cloud Analyst Summit/ Data & AI/ July 2019 / © 2019 IBM Corporation
Is AI your priority? Start with a data strategy
The union between IBM and Intel is supercharging the ability of data scientists to drive better insight and better business outcomes in a way that has never been seen before.
Intel and IBM are great partners and closely aligned in becoming more data-centric.
Intel's participation and contribution is meaningful because customers can run Cloud Pak for Data at speed on their Intel-based infrastructure
Data and AI / © 2019 IBM Corporation
“ Include a quote if one is available, we can always add this at closeout. ”
-- Include quote author’s name and title
“ Include a quote if one is available, we can always add this at closeout. ”
-- Include quote author’s name and title
Solution
Cloud Pak for Data
Read the blog and watch the videoModernize
Unlock the value of your data for an AI and multicloud world
Melvin Greer
Senior Principal Engineer and
Chief Data Scientist - Americas,
Intel Corporation
IBM Cloud Pak for Data is really important because it helps to do a couple of things that are mind blowing for data scientists — auto discovery of data and rapid integration of hyper relevant data.
Industry: TechnologyGeography: North America
Intel
Client Value PropositionWhy ?
Data & security regulation requirements around data governance, data privacy, and security– New data protection regulations around the world GDPR, etc.– Increasing focus on IT security requirements
Faster data access for analytics and Line of Business– Average customer has 168 difference data sources they want
to use within analytics– 81% of business are having issues preparing data required for
AI
AI trust and explainability– 85% of companies see AI as a strategic opportunity– 60% of companies see regulatory constraints as barriers to
implementing AI
Common experience across hybrid clouds; provision containerized data and AI services in minutes behind your firewall, instead of taking weeks– Only 20% of workloads have moved to the cloud– 75% of enterprises will modernize existing applications over
the next 3 years
IBM Differentiators
Data Governance– Data privacy and governance by design: data discovery and curation, with
policy and rules management – Metadata management and shopping for data– Smarter compliance: Regulatory ML, Accelerators, FISMA HIGH
certification, etc.
Data Virtualization– Query all of your data sources as one– Governance, security, and scalability by design– 5X faster data access; 40X faster than federation
Governing & Operationalizing AI– Governed AI lifecycle management
– Quality-of-Service optimization– AI model trust and explainability
Delivery Models– OpenShift enables consistent hybrid deployment patterns with any cloud
(public, private or on-premises)– Hyper-converged System offering combines storage, compute,
networking, and software into a single system to reduce complexity and increase scalability
– Targeted professional services offerings to accelerate use case execution29
Summary
• Optimized for fast & flexible data service lifecycle management
• Portability across on-premises, private clouds & public clouds
• Appliance available
Multi-Cloud
• One experience across all data services
• Coherent, efficient, and scalable data services experience
• Ease of provisioning, monitoring, and moreUnified Console
• IBM add-ons
• Open Ecosystem
• Data Virtualization
Ecosystem & Virtualization
• Data automatically integrated with governance capabilities for catalog and search subject to policies & rules – enabling self-service data discovery
Data Governance & Self-Service Analytics
IBM Cloud Pak for Data allows:
30
Experience It ! https://www.ibm.com/cloud/garage/cloud-pak-experiences-for-data/
Recommended