What's new in IBM Cloud Pak for Data?

• Watson Studio
• Watson Machine Learning
• Jupyter Notebooks with Python

Python 3.7 replaces Python 3.6

Python 3.6 is being deprecated and replaced by Python 3.7. (However, you can continue to use Python 3.6 environments.)

For details on using Python frameworks with machine learning models and deployments, see Supported frameworks.

To migrate to Python 3.7 in Jupyter Notebooks with Python, install the appropriate service for your environment:

• Jupyter Notebooks with Python 3.7 (runtime-addon-py37)
• Jupyter Notebooks with Python 3.7 for GPU (runtime-addon-py37gpu)

For details, see Installing Jupyter Notebooks with Python.

For details, see Installing IBM Cloud Pak for Data.

If you have a previous version of the service installed, you must upgrade to the 3.2.2 version of the service before you upgrade your OpenShift installation. For links to installation and upgrade instructions, see Cognos Analytics.

For links to installation and upgrade instructions, see Cognos Dashboards.

For links to installation and upgrade instructions, see Data Virtualization.

For links to installation and upgrade instructions, see DataStage.

If you have a previous version of the service installed, you must upgrade to the 3.0.2 version of the service before you upgrade your OpenShift installation. For links to installation and upgrade instructions, see Db2.

For links to installation and upgrade instructions, see Db2 Big SQL.

For links to installation and upgrade instructions, see Db2 Data Gate.

If you have a previous version of the service installed, you must upgrade to the 3.0.2 version of the service before you upgrade your OpenShift installation. For links to installation and upgrade instructions, see Db2 Event Store.

For links to installation and upgrade instructions, see Db2 for z/OS Connector.

If you have a previous version of the service installed, you must upgrade to the 3.0.2 version of the service before you upgrade your OpenShift installation. For links to installation and upgrade instructions, see Db2 Warehouse.

For links to installation and upgrade instructions, see Decision Optimization.

For links to installation and upgrade instructions, see Execution Engine for Apache Hadoop.

If you are installing Financial Crimes Insight on OpenShift 4.5, you must use the updated images. For details, see Financial Crimes Insight platform installation steps.

For links to installation and upgrade instructions, see Guardium External S-TAP.

For links to installation and upgrade instructions, see Master Data Connect.

If you have a previous version of the service installed, you must upgrade to the 3.0.2 version of the service before you upgrade your OpenShift installation. For links to installation and upgrade instructions, see MongoDB.

For links to installation and upgrade instructions, see Open Source Management.

If you have a previous version of the service installed, you must upgrade to the 3.0.1 version of the service before you upgrade your OpenShift installation. For links to installation and upgrade instructions, see Planning Analytics.

For links to installation and upgrade instructions, see Regulatory Accelerator.

For links to installation and upgrade instructions, see RStudio Server with R 3.6.

If you have a previous version of the service installed, you must upgrade to the 3.0.2 version of the service before you upgrade your OpenShift installation. For links to installation and upgrade instructions, see SPSS Modeler.

For links to installation and upgrade instructions, see Streams.

For links to installation and upgrade instructions, see Watson AIOps.

For links to installation and upgrade instructions, see Watson Assistant.

For links to installation and upgrade instructions, see Watson Assistant for Voice Interaction.

For links to installation and upgrade instructions, see Watson Discovery.

For links to installation and upgrade instructions, see Watson Knowledge Catalog.

For links to installation and upgrade instructions, see Watson Knowledge Studio.

For links to installation and upgrade instructions, see Watson Language Translator.

• For x86-64: cpd-3.0.1-wml-patch-3
• For POWER: cpd-3.0.1-wml-patch-2

For links to installation and upgrade instructions, see Watson OpenScale.

For links to installation and upgrade instructions, see Watson Speech to Text.

For links to installation and upgrade instructions, see Watson Text to Speech.

• The getting started experience has been merged with the home page so that you have a single point of entry to the platform.

  The links that you need to get up and running are at the top of the page. The links help you identify common tasks that are associated with your role. If you don't need the links, you can easily hide them so that you can see more of the cards on the home page.

  
• The home page includes more useful cards that give you access to your notifications, recent projects, recent analytics dashboards, requests, and more.
  
  

  The cards that are displayed depend on the services that are installed on top of Cloud Pak for Data.

• The links you need, at your fingertips.
  The home page includes quick links to commonly used features, such as projects, catalogs, analytics dashboards, and more.

  

  The links that are displayed depend on the services that are installed on top of Cloud Pak for Data.

For details, see Creating custom cards.

Carbon makes it easier for you to know exactly how the web client will behave. With Carbon, you can spend less time learning how to use the platform and more time putting the services to use to accelerate your business.

• Specify a custom name to display in the web client
• Adding your own logo to the home page

For details, see Customizing the branding of the web client.

Many of the data sets are included with your purchase of Cloud Pak for Data. However, some data sets are separately priced.

You can use the external data to expand the capabilities of your models and applications. For details, see External data sets.

• Red Hat OpenShift Container Storage
• IBM Cloud File Storage

For information on the types of storage that are supported with Cloud Pak for Data, see Storage considerations.

In addition, many services include support for additional storage types. For information about which storage types are supported for each service, see System requirements for services.

For details, see Auditing Cloud Pak for Data.

For details, see Integrating with Cloud Platform Common Services.

If you give a user all of these permissions, it is equivalent to giving the user the Administer platform permission.

For details, see Permissions.

For details, see Backup and disaster recovery.

The documentation also includes information on how to plan for disaster recovery.

• Brazilian Portuguese
• French
• German
• Italian
• Japanese
• Simplified Chinese
• Spanish
• Traditional Chinese

Several services are also available in Russian and Korean.

For details, see Language support.

Enterprise Edition

The IBM Cloud Pak for Data Enterprise Edition package includes non-production parts that you can install in your development, test, or quality assurance environments. The non-production Enterprise Edition parts make it easier to transition workloads to your production environments.

Standard Edition

IBM Cloud Pak for Data Cloud Native Edition has been renamed to IBM Cloud Pak for Data Standard Edition. As with Cloud Native Edition, Standard Edition places limits on the number of virtual processor cores (VPCs) that you can have in your cluster.

New visualization charts in Data Refinery

Data Refinery introduces six new charts. To access the charts, click the Visualizations tab in Data Refinery, and then select the columns to visualize. The chart automatically updates as you refine the data.

• Bubble charts display each category in the groups as a bubble.
• Circle packing charts display hierarchical data as a set of nested areas.
• Multi-charts display up to four combinations of Bar, Line, Pie, and Scatter plot charts.

  You can show the same kind of chart more than once with different data. For example, two pie charts with data from different columns.

• Radar charts integrate three or more quantitative variables that are represented on axes (radii) into a single radial figure.

  Data is plotted on each axis and joined to adjacent axes by connecting lines. Radar charts are useful to show correlations and compare categorized data.

• Theme river charts use a specialized flow graph that shows changes over time.
• Time plot charts illustrate data points at successive intervals of time.

For details, see Visualizing your data.

Backup, restore, and disaster recovery

It is a best practice to back up the Data Virtualization service. You can use the cpdbr utility to back up and restore all persistent volumes that are associated with your Data Virtualization installation. The cpdbr utility performs local volume snapshots and restores the service from a local snapshot via Portworx.

For details, see Backing up and restoring Data Virtualization.

Support for additional data sources

You can connect to the following data sources in Data Virtualization:

• Google BigQuery
• SAP HANA
• Amazon Redshift
• Netezza®

For details, see Adding data sources (Data Virtualization).

Virtualize data from remote TSV files

Virtualization is now available for data that is stored in TSV files on remote data sources.

For details, see Creating virtualized tables from files (Data Virtualization).

Governance enhancements for virtual objects

Data Virtualization uses policies and business terms in Watson Knowledge Catalog to govern your virtual data. These policies consist of data protection rules with conditions that use business terms, tags, data classes etc. For example, you can use data protection rules to deny access to a virtual table or to mask data in columns of a virtual table.

Additionally, you can publish your virtual objects to Watson Knowledge Catalog.

For details, see Governing virtual data (Data Virtualization).

Improve query performance by using multiple worker nodes

You can configure the Data Virtualization service to run on multiple worker nodes. Running the service on multiple worker nodes improves query performance and enhances parallelism for query processing.

When you provision Data Virtualization, you can specify the number of worker nodes to allocate to the service.

For details, see Enabling query processing parallelism on multiple worker nodes (Data Virtualization).

Reduced footprint to provision the service

Data Virtualization has a reduced footprint when you initially provision the service.. This new footprint reduces default minimum requirements and enables a more efficient usage of service resources.

For details, see Preparing to install the service (Data Virtualization).

DataStage Enterprise and DataStage Enterprise Plus

You can choose from two different DataStage offerings: DataStage Enterprise and DataStage Enterprise Plus.

Use DataStage Enterprise for all the benefits of classic DataStage, designing and running data flows that move and transform data. Use DataStage Enterprise Plus to get all the capabilities of DataStage Enterprise with additional features for data quality. These features include:

• Cleansing data by identifying potential anomalies and metadata discrepancies.
• Identifying duplicates by using data matching and probabilistic matching of data entities between two data sets.

Automatic job generation

You can use a new job generation feature that automates and simplifies data movement, taking data from a source (X) and moving it to a target (Y). Rather than having to craft a job to move this data by using the DataStage Edition canvas and palette, and then dragging connectors and stages to build a job, you can use the new job template to follow a simple series of steps and generate a parallel job. You can then use this parallel job to move the data. Also, if you are an administrator you can define target rule sets that define best practices and have users apply those to their job templates and generated jobs.

For details, see Generating a job by using a template.

Edit column metadata

You can define and edit column metadata. For example, for a column, you can specify which delimiter character should separate text strings or you can set null field values. With the ability to edit column metadata you get more fine-tuned processing of your jobs and more useful data transformation.

For details, see Running a data transformation job.

Shared containers support

You can create a container that has its own job flow on the canvas and share it with other jobs.

For details, see Running a data transformation job.

Connectors

• SAP OData
• z/OS file

For details, see Supported connectors.

Stages

• You can use the following stages:
  • Complex flat file
  • Slowly changing dimension for warehousing
• Transformer stage enhanced to support Slowly changing dimension stage
• Transformer stage supports tabs such as build, surrogate key, and triggers.
• All stages support property tabs General, Stage Advanced, and Output Advanced.

For details, see Supported stages.

Enhancements for the Hierarchical stage

• You can use the following steps in the Hierarchical stage:
  • JSON composer step
  • JSON parser step
  • REST step
  • Test assembly
  • Details inspector
  • Create/view contract libraries
  • Administration
• Support for 10 in-build operator steps.
• Tree-based view

For details, see Supported stages.

Upgraded optimization engines

CPLEX® V12.9 and V12.10 optimization engines are now available. When you click the Run model button in the Decision Optimization model builder, CPLEX V12.10 is used to solve the model.

More data formats

You can now import data in other formats (for example XLS, JSON and so on), as well as CSV files or connected files, in the Prepare data view of the Decision Optimization builder.

For details, see Prepare data view.

Customize the deployment runtime

The Decision Optimization runtime used in deployment can be extended to include other APIs.

For details, see ExtendWMLSoftwareSpec notebooks in the DO-samples repository on the Decision Optimization GitHub site.

Exported notebooks contain all files

When you export a notebook from the Decision Optimization model builder from a model containing multiple files, the additional files are also exported with the model.

For details, see the information on the Generate a notebook from a scenario option in Scenario panel.

Support for new storage types

The Db2 service supports additional types of storage.

For details, see System requirements for services.

Separate storage types for metadata and user data

The Db2 service now uses different storage classes for metadata and user data.

For details, see Configuring database storage for Db2.

Persistent volume claim templates

You can define a volumeClaimTemplate so that a new persistent volume claim is automatically created for each replica.

For details, see Configuring database storage for Db2.

Automated failover for HADR

The Db2 service includes support for automated failover of Db2 high availability disaster recovery (HADR) by using a mechanism called a governor.

Also, during a new deployment of the primary and standby HADR databases with the web console, you can specify that the service endpoints that are required to transport HADR traffic are automatically created as part of the deployment.

For details, see Db2 high availability disaster recovery (HADR).

Support for new storage types

The Db2 Warehouse service supports additional types of storage. The types of storage that are available depend on whether you are deploying with the symmetric multiprocessing (SMP) architecture or the massively parallel processing (MPP) architecture.

For details, see System requirements for services.

Separate storage types for metadata and user data

The Db2 Warehouse service now uses different storage classes for metadata and user data.

For details, see Configuring database storage for Db2 Warehouse.

Persistent volume claim templates

You can define a volumeClaimTemplate so that a new persistent volume claim is automatically created for each replica.

For details, see Configuring database storage for Db2 Warehouse.

New connection type

You can use the new Impala via Execution Engine for Hadoop connection to browse, preview, and refine data in Impala tables.

R scripts

You can schedule jobs for R scripts on remote Hadoop clusters using Execution Engine for Hadoop. For details, see R scripts.

Customizing Spark runtime settings

Users can now fine tune their Spark session using user-defined session variables. The Hadoop administrator must first determine what options to modify and the default values or ranges that are allowed. Once the system is set up, users can then create an environment with these options to run their jobs or launch their notebook. For more details, see Hadoop environments and Configuring the Hadoop integration with Cloud Pak for Data.

Updated CDH support

Execution Engine for Hadoop is now supported on CDH version 6.1, 6.2, and 6.3

• Hadoop 3.0 support.
• Transaction List Screening multi-pipeline support.
• New performance visualization.
• Narrative Generation Capabilities to allow customizable templates and point-in-time data report.
• Enhancements to Policy Admin tool to support nested policies.
• Voice processing using Watson Speech to Text.

Understand your open source landscape

The Open Source Management includes a dashboard that gives you insight into how you use open source packages across your organization. For example, the dashboard lets you see:

• The number of projects that use open source packages
• The number of packages that use any given type of open source license
• The number of packages that are affected by vulnerabilities
• The number of times users interact with open source packages
• And so much more

For details, Open Source Management.

Git integration to collaborate on R Shiny® apps and R scripts

You can configure RStudio to use a Git repository, which enables you to collaborate on R Shiny apps and R scripts in RStudio. When you are ready to use your apps and scripts, you can pull them from the Git repository and add them to an analytics project. The apps and scripts become project assets, which you can deploy by promoting them to deployment spaces.

For details, see Analyzing data with RStudio.

R scripts

You can create R scripts in RStudio from an analytics project that is integrated with a Git repository. You can schedule jobs for R scripts on local and remote Hadoop clusters and pass parameters for running the jobs. You can also include R scripts in a project export.

For details, see R scripts.

Support for SAV files

You can now import or export SPSS Statistics SAV files from SPSS Modeler.

New cross-validation options

New cross-validation options are available in the Auto Classifier node and the Auto Numeric node.

Reduced footprint to provision the service

Streams has a reduced footprint when you initially provision the service, resulting in decreased costs because of reduced vCPU, RAM, and disk requirements.

For details, see System requirements for services.

Streams flows code

For users who want to understand and enhance Streams flows Python source code, flows are now available for download and export to notebook, where they can be run, tweaked, and expanded with advanced API features. Look for the download code and export to notebook buttons on the flow editor toolbar.

Speech To Text Gateway

Streaming data applications can ingest audio data and transcribe into text by using the Watson Speech to Text service. Audio data is either read from files for batch workloads or streamed through a telephony infrastructure for real-time workloads, such as call center support.

Integration with Watson Machine Learning

Streaming data applications can use the online scoring functionality of Watson Machine Learning.

Watson AIOps 2.0.0 is now composed of the following components: AI Manager, Metric Manager, Event Manager, and Topology.

AI Manager for Watson AIOps 2.0.0 now supports application groups. Application groups are a means for isolating streams of data from one another. You can now group multiple SRE groups alongside multiple internal clients (for example, several business units) all within a single cluster that runs Watson AIOps 2.0. Not only does it simplify your IT footprint, but it also significantly reduces your IT costs because you need to provision only a single instance to share across many application groups.

Event grouping

• Data validation before log anomaly training ensures that your training is not affected by quality issues in your input data sets.
• Template-based clustering adds to the accuracy of establishing patterns by using detailed entity extraction information from log contents. Those patterns are then applied when AI Manager provides a holistic view of the problem through a story.
• Trained models can now be verified and evaluated at training time and you can save the logs from evaluating your models. By evaluating your models, you can gain a higher degree of confidence in your models before your deploy them.
• Stories can now include anomalies over a range of time.

Topology and localization

• Legacy topologies are now handled as part of the hybrid approach to topology management. Legacy topologies are integrated with Kubernetes topologies.
• Localization and blast radius for applications across the hybrid environments are now supported.
• Localization and blast radius now support using historical topology as it existed at the time of the story and are available from your reported stories directly in Slack. You can now jump straight in to an interactive IBM Netcool Agile Service Manager dashboard to view where events occurred and the scope of their impact right from Slack.
• Log readability has been improved.

More accurate automatic term assignments during advanced data curation

Now, if you reject term assignments that are generated by the class-based and linguistic name matching services, they are also excluded when repeating the same data analysis. When you reject terms that the machine learning service provides, it learns from your actions, and generates better results in future.

For details, see Automatic term assignment (Watson Knowledge Catalog).

Enhanced data analysis during advanced data curation

• You can speed up relationship analysis and overlap analysis by filtering out columns by name, column type, or first N candidates.
• The delta row in the data quality violations table shows how the number of violations changed between the last two analyses. You can now view trends in data quality for each data asset. A user can quickly visualize how their Data Quality has changed over time by data asset, based on their specified time interval.

Supported data sources and connections

• Teradata connections and connected data assets are now synchronized from the Information assets view to the default catalog.
• Information assets for files are now synced to default catalog.
• You can now create connection assets in catalogs to the following data sources:
  • Impala via Execution Engine for Hadoop
  • SAP OData
  • Planning Analytics (formerly known as TM1®)
• You can now supply personal credentials for connection assets and connected data assets that require them.

Data protection rules are more powerful and flexible

You can now include classifications in criteria when you create data protection rules.

For details, see Managing data protection rules.

Migrate assets from IBM InfoSphere Information Server

You can migrate assets from IBM InfoSphere Information Server versions 11.7.1.x to the Watson Knowledge Catalog service on Cloud Pak for Data Version 3.0.1.

For details, see Migrating data from IBM InfoSphere Information Server to IBM Cloud Pak for Data (Watson Knowledge Catalog).

AutoAI improvements

• AutoAI includes more powerful visualizations, as well as support for larger data sources and more estimators. Additionally, you can now create a batch deployment for an AutoAI model.

  For details, see AutoAI overview.

• Tech preview You save an AutoAI experiment pipeline as an auto-generated notebook so that you can audit the transformations that created the pipeline. The documentation in the notebook and in the autoai-lib reference helps you interpret the results.

  For details, see Saving an AutoAI generated notebook.

More batch deployment options

• Batch deployments have been significantly enhanced to allow for the programmatic connection to an expanded set of data sources. For example, connect to Db2 for an SPSS deployment.
• Create a batch job that accepts multiple inputs from Db2 or DashDB for an SPSS deployment.
• You can create batch deployment jobs so you can schedule runs or change the runtime environment for a deployment.

For details, see Batch deployment details.

More deployment options

• New deployment support for a wider variety of frameworks and assets helps you deploy assets your way.

  For details, see Supported frameworks.

• Update an online deployment with a better performing model from an existing endpoint.

  For details, see Creating an online deployment.

• You can promote data connections as well as data assets for use with deployments and scripts.

  For details, see Deploying assets.

• You can deploy Python scripts and R Shiny apps to further extend your deployment and model management options.

  For details, see Deploying assets.

More import and export options

• New import and export capabilities make it easier for you to share deployment spaces and deployed content. You can export a space and all of its assets to a file or import a space and all of its assets from a file.
• Share models more easily with the capability of importing a model in PMML format.

For details, see Analytics deployment spaces.

Context-sensitive help pane

The context-sensitive help pane provides help that is based on what you are doing. The results are displayed inside the user interface for added convenience.

Model risk management

• Integrates with IBM OpenPages governance, risk, and compliance platform
• Supports pre-production model environments
• Ability to compare two models

Customization options for accessing custom libraries

You can customize the software configuration of Jupyter notebook environment definitions in Watson Studio by using conda channels and pip to access custom packages and libraries. If access to the public network is not available or desired for security reasons, you can customize the conda and pip configuration to access libraries by alternate methods, or you can build your own custom images.

For details, see Customizing environments.

Python Scripts

When working in JupyterLab, you can now share your work in Python scripts as well as in notebooks. You can collaborate through the Git repository associated with your project and later pull changes back to the project to create a script asset. You can schedule jobs for scripts and pass parameters for running the jobs. You can promote scripts to deployment spaces and include them in a project export.

For details, see JupyterLab.

Supported data sources and connections

• Impala via Execution Engine for Hadoop
• OData
• Planning Analytics (formerly known as TM1)
• SAP OData

Many of the services that supplement Watson Studio also have new features. For details, see the rows for those services.

A good plan starts with good data. Ensure that your plans are based on data from across your business with IBM Planning Analytics powered by TM1.

Planning Analytics is an AI-infused solution that pulls data from multiple sources and automates the creation of plans, budgets, and forecasts. Planning Analytics integrates with Microsoft Excel so that you can continue to use a familiar interface while moving beyond the traditional limits of a spreadsheet. Infuse your spreadsheets with more analytical power to build sophisticated, multidimensional models that help you create more reliable plans and forecasts.

IBM Guardium External S-TAP is a component of Guardium that works with databases that are hosted on Cloud Pak for Data. The service provides compliance monitoring and data security.

You can install and configure the External S-TAP service in high-availability mode to intercept TCP/IP traffic (plain-text or encrypted) between Cloud Pak for Data users and database services. The intercepted traffic is sent to the Guardium collector for parsing, policy enforcement, logging, and reporting.

To use the External S-TAP service, you must be entitled to use IBM Guardium Data Protection.

Power® your business applications with trusted master data.

Master data is the high-value, core information that supports critical business processes across your enterprise. Master data is at the heart of every business transaction, application, and decision.

The Master Data Connect service uses a RESTful API to provide geographically distributed users with fast, scalable, and concurrent access to your organization's most trusted master data from IBM InfoSphere MDM. By making your trusted master data available to business applications, you can capitalize on the benefits that master data brings.

Users and systems can use the Master Data Connect API to access and search master data, enabling your mobile and online applications to access trusted master data in a timely and efficient way. For example, you can improve your sales processes by using Master Data Connect as a real-time provider of master data for Salesforce.com.

Use standard SQL to query your data on Hadoop or Object Stores with Db2 Big SQL

Db2 Big SQL is an advanced query service that makes it easy to analyze data in object stores or Hadoop using ANSI SQL that is optimized for advanced analytics in big data environments. Powerful Db2 open source technologies are the prime driver for machine learning, interactive, ad hoc, and batch analytics use cases on open source file formats stored on Hadoop and object stores.

With data sizes ranging from gigabytes to petabytes, business analysts or data scientists run interactive queries to explore and understand data before building models or charts. With its robust scalability and performance, Db2 Big SQL empowers users and applications to unlock insights from data with the analytics tools of your choice, while achieving high concurrency for business intelligence workloads by running complex queries more efficiently.

Each virtual database copy can be refreshed to any point in time in a matter of minutes and can be masked to protect sensitive data. In addition, virtual database copies use almost no storage, so you save on storage costs.

Give your users access to production data without impacting priority workloads or compromising data security and privacy. Get started with the Virtual Data Pipeline service to accelerate your analytics and modernize your applications.

Data scientists in analytics projects can create custom environment definitions that include the conda channels and packages from the repository and then use those environments to run Jupyter notebooks and scripts.

With the Anaconda Repository with IBM Cloud Pak for Data service, you can access more than 7,500 open-source packages (Conda-Forge, CRAN, PyPI) from your central enterprise repository and add your own proprietary packages. Get Conda package updates in real time, as they are released.

Block, exclude, and include packages according to your enterprise standards. Control which packages your team can download and who can access them. Keep vulnerabilities and unreliable software out of your data science and machine learning pipeline and manage dependent packages.

For details, see Demand planning on the Cloud Pak for Data Community.

For details, see Manufacturing Analytics with Weather on the Cloud Pak for Data Community.

For details, see Retail Predictive Analytics with Weather (using SPSS and Cognos) on the Cloud Pak for Data Community.

For details, see Sales Prediction using The Weather Company Data on the Cloud Pak for Data Community.

For details, see Telco churn on the Cloud Pak for Data Community.

For details, see Utilities Customer Attrition Prediction on the Cloud Pak for Data Community.

For details, see Utilities Customer Micro Segmentation on the Cloud Pak for Data Community.

For details, see Utilities Demand Response Program Propensity on the Cloud Pak for Data Community.

For details, see Utilities Payment Risk Prediction on the Cloud Pak for Data Community.

For more information, see the description of the new Planning Analytics service.

For more information, see the description of the new Virtual Data Pipeline service.

• Red Hat OpenShift Container Platform Version 3.11
• Red Hat OpenShift Container Platform Version 4.3

For more information about Red Hat OpenShift Container Platform, see System requirements for IBM Cloud Pak for Data.

If you are currently running Cloud Pak for Data Version 2.5 on Red Hat OpenShift Container Platform Version 3.11 and want to migrate to Cloud Pak for Data Version 3.0.1 on Red Hat OpenShift Container Platform Version 4.3, see Migrating Cloud Pak for Data data from Red Hat OpenShift Version 3.11 to Version 4.5.

For a list of the services that you can install on IBM POWER hardware, see hardware requirements in the System requirements for services topic.

• For information on upgrading the control plane, see Upgrading IBM Cloud Pak for Data control plane.
• For information on upgrading services, see the appropriate section under Services in the catalog or Services outside the catalog.

  Some services use a different mechanism for upgrade and some services do not currently support upgrade. See the documentation for the services that you currently have installed to determine what is supported for each service.

Because this feature was deprecated, the related feature that allowed you to set your target usage was also deprecated.

You can no longer create connections to Object Storage Open Stack Swift (Infrastructure) from Watson Studio or Watson Knowledge Catalog. This type of connection is deprecated in Cloud Pak for Data Version 3.0.1.

If your project contains a connection to Object Storage Open Stack Swift (Infrastructure), the connection will no longer work.

The Python and SPSS operators are no longer supported in Streams Flows. The WML Deployment operator replaces both of these operators.

For details on fixing flows that contain these operators, see Troubleshooting a streams flow.