Predix Insight

Overview
Bangalore

Feb 2014
Agenda
• What is Predix Insight
• Use Case
• Predix Insight value proposition
• Customer pain points/requirements
• Predix Insight 1.0 value proposition
• Feature Set
• Roadmap
• Predix Insight Contacts
• Demo

1 GE Internal
Predix
The software platform for the Industrial Internet
Predictivity Energy Aviation Transportation Healthcare
Solutions

People at Work Machine Server

Grid/Utility

Network
On Premise
Wind Farm
or
Turbine
Cloud

Predix Predix Server

Experience Predix Predix Predix
Machine Predix Net Asset Insight
Predix Predix
Reader Cards
Predix Core
Predix Micro
Predix Predix
V Connect Predix Data Fabric

2 GE Internal
What is Predix Insight?

Predix Insight is a scalable framework to

rapidly and iteratively compose cross-
language, cross-platform analytic applications
upon high-volume, high-variety, and high-
velocity industrial data.

Predix Insight provides a contribution platform

for analytic applications.

3 GE Internal
Predix Insight Main Use Cases

Business use case: refer to the wiki

https://devcloud.swcoe.ge.com/devspace/display/PHMCAF

4 GE Internal
Insight Use Case: Engine Diagnostic

Data Network Data Center

Diagnostic
Results

Notifications
to Airline Aviation Operation Center
Data Connector

Predix Insight
Objective:
To monitor and analyze the Load Balancer Load Balancer
health of engines to increase
coverage, detection and lead
Diagnostic Diagnostic Diagnostic Diagnostic
time and reduce false alarms Analytic Analytic Analytic Analytic
across the fleets

5 GE Internal
Customer Pain Points/Requirements
Audience: Solution Developers, Data Scientists and
Operations

① Slow time to market with limited differentiation in

developing analytics solutions

② High TCO to build analytics solutions

③ Limited collaboration in analytics development

and operations to foster innovation

6 GE Internal
Predix Insight Value Proposition
Audience: Solution Developers, Data Scientists and
Operations

① Fast time to market to develop analytics solutions

① Low TCO to build analytics solutions

② Maximize collaboration in analytics development

and operations to foster innovation

7 GE Internal
Predix Insight Operational Overview

8 GE Internal
Predix Insight Capabilities
Persona Category Capabilities
Data Scientist Content • Analytic Catalog of algorithms
• Generalized Metadata for algorithms
Solution Tools • Web based Orchestration Editor
Development • User guided analytic configuration
• Simplified cross-language data binding
• Standardized orchestrations based on BPMN 2.0
• Asset based data binding strategy framework
Operations Runtime • Multi language analytic support (eg: Java, C, C++, C#)
Execution • Intermediate data caching
• In-memory analytic execution
• Long running analytic execution
• Configurable horizontal scaling
• Framework for supporting variety of data
(aggregation of Asset vs. Fleet vs. Network data)
Operation • Analytic Configurations migration and ease of
Support deployment
• Analytic Monitoring Dashboard
• Condition based Dispatch

9 GE Internal
Roadmap Current Features Future Roadmap

Predix Insight
Tools Analytic Rendering & Visualization Operation Support

Data Graph,
Analytic Alarm / Condition Analytic
Ingestion Test Trending Decision Analytic
Orchestion Event / based Lifecycle
and Module & Support Monitoring
Editor Rules Dispatch Mgmt
Exploration Reporting

Analytic Catalog
Content

Machine Statistical and Data Condition

Math Manipulation Prognostic Predictive
Learning Monitoring
Analytics Analytics Analytics Analytics
Analytics Analytics

Analytic Orchestration Engine

Execution
Runtime

Small BigData Analytics Instance Big

Streaming
In-Memory (eg: Map Reduce) In-Memory Grid
Common

Access
Data

Data Reader/Writer
Predix Asset Entity Models Field Metadata
Connector

Predix Core
10 GE Confidential
GE Internal
Predix
Q1 CY2014 Q2 CY2014 Q3 CY2014 Q4 CY2014

Predix Core 14.1 Predix Core 14.2 Predix Core 14.3

Predix DataFabric Predix DataFabric

14.2 14.3

Predix Machine 14.2 Predix Machine 14.3

Predix Net 14.2 Predix Net 14.3

Predix Asset 14.1 Predix Asset 14.2 Predix Asset 14.3

Predix Insight 14.1 Predix Insight 14.2 Predix Insight 14.3

Predix Experience Predix Experience Predix Experience

14.1 14.2 14.3

This represents GE Software’s current view of its product development cycle and future directions. It is intended for information purposes only, and should not be interpreted as a commitment on
the part of GE Software. GE Software makes no warranties, express or implied, in this document.

11 GE GE Internal – Distribution authorized to individuals with need to know only

Confidential Software Technology Conference | GESoftware.com
11
Predix Insight
Q4 2014
Q1 2014 Q2 2014 (May) Q3 2014 (August)
(December)
Predix Insight 14.1 Predix Insight 14.2 Predix Insight 14.3

Features Features Features

• Name Value Selection Filter • Asset – Insight Integration • Analytic LifeCycle
• Simplification of Installation • Monitoring Dashboard Management
and Wrapping Analytics • Multi-Language Support for R • Data Exploration (leverage
• Platform alignment with and Python Core/Visualization tools)
Hibernate replacement by • Hadoop / MapReduce support • Orchestration Sub-processes
Open JPA for Analytics • Security Enhancements –
• UI modernization and Ext JS • Install and Packaging RBAC etc.
replacement Enhancements
• Clustering / Multi-Node • Enrichment of Analytics Benefits
Topology Support Catalog • Facilitate improved sharing of
• Design Collaboration / analytics and analytical
planning for Hadoop and Asset Benefits solution components across
integration • Leverage Asset metadata from GE businesses
Predix Asset for quick • Support for Full analytic
Benefits deployment of Analytics lifecycle management
• Improved flexibility for data solutions • Security and role based user
selection for analytics • Reference application for M&D management
• Faster install and deployment of Assets
of analytics solutions • Support for majority of
• Modern UI based on UX best Analytics platforms in use in
practices GE

This represents GE Software’s current view of its product development cycle and future directions. It is intended for information purposes only, and should not be interpreted as a commitment on
the part of GE Software. GE Software makes no warranties, express or implied, in this document.

12 GE GE Internal – Distribution authorized to individuals with need to know only

Confidential Software Technology Conference | GESoftware.com
12
Contacts
Predix Insight
• Product Management
• Ibrahim Gokcen, [email protected]
• Shyam Nath, [email protected]
• Software Development Manager
• Nagesh Nayudu, [email protected]
• Product Marketing
• Venkat Eswara, [email protected]
• Solution Services Team
• [email protected]

13 GE Internal
 Thank You!
Questions and suggestion are always welcome.

14 GE Internal
 Backup
Insight 1.0 Messaging

15 GE Internal
 Demo

16 GE Internal
Customer Pain Points/Requirements
Audience: Solution Developers, Data Scientists and
Operations
Slow time to market with Limited collaboration in
limited differentiation in High TCO to build analytics development
developing analytics analytics solutions and operations to foster
solutions innovation
• Longer time to develop and • High cost for development and • Inefficient time and use of data
productize analytic solutions - maintenance towards specialized scientists in dealing with
takes months to years skills needed in advanced data disjointed, multiple analytical
analytics language environment vs.
• Limits scale, minimizes innovation focusing on high-value business
and increases time to • Expensive and longer customer problems for the customers
operationalize feedback cycle time for point
solutions • Lack of common place to share
• Limited extensibility with point analytics, algorithm
solution approach to aggregate • High cost and risk with the
from asset to fleet to network level introduction of new technologies • Inability to manage and distribute
and programming languages secured data sets in a centralized
location

17 GE Internal
Predix Insight Value Proposition
Audience: Solution Developers, Data Scientists and
Operations
Maximize collaboration in
Fast time to market to
Low TCO to build analytics development
develop analytics
analytics solutions and operations to foster
solutions
innovation
• Rapid development of analytics • Lower development and • Increase efficiency by focusing on
solutions using Predix Insight - maintenance costs with high value business problems
months to weeks collaboration and common
frameworks • One common environment to
• Scalable common architecture for share algorithms
analytics • Minimize cost and cycle time to
incorporate customer • Ability to manage and distribute
• Extensible framework to requirements with standardized, secured data sets in a centralized
aggregate at asset, fleet and collaborative analytic environment location
network level
• Minimize costs and risks
associated with the introduction of
new technologies and
programming languages

18 GE Internal
① Fast time to market to develop analytics
solutions
Pain Point 1: Value Proposition 1:

Slow time to market with limited differentiation in developing Fast time to market to build analytics solutions
analytics solution
1.Longer time to i. variety of disjointed analytic tools in GE from 1. Rapid development i. one place to consume analytic tools from Excel,
develop and Excel, Mathlab, Java and C++ of analytical Mathlab, Java and C++
productize analytic ii. lack of standard way to develop and integrate solutions using ii. one standard way to combine analytics
solutions - takes analytic solutions Predix Insight – workflows written in different languages
iii. significant development effort for new analytic months to weeks iii. ability to reuse services and workflows
months to years
solutions due to lack of reuse of existing minimizes time to develop new analytic
solutions solutions

2.Limits scale, i. inability to leverage analytic solutions across 2.Scalable common i. reuse common themes such as monitoring and
minimizes innovation businesses, as current analytic solutions are architecture for diagnostics across distinct businesses
and increases time to siloed in a given business group built on a given ii. a standardized configurable fashion to map
analytics
technology inputs and outputs to analytics which minimizes
operationalize
ii. significant effort to integrate and operationalize integration effort
each analytic solution iii. independently scalable layered architecture to
iii. limited scalability built on siloed infrastructure scale horizontally

3.Limited extensibility i. significant custom coding needed to combine 3.Extensible i. orchestration of analytics and ability to quickly
and/or orchestrate multiple analytics framework to assimilate new functionality with reusable
with point solution
ii. lack of standard API's to reuse and built upon aggregate at building blocks
approach to ii. standardized wrappers and data interface for
the current analytics asset, fleet and
aggregate from asset analytics
iii. inability to scale horizontally as the current network level
to fleet to network iii. decoupled scalable architecture for horizontal
analytics are built to one infrastructure identity
level only scalability at each tier of the analytic solution

19 GE Internal
 ② Low TCO to build analytics solutions
Pain Point 2: Value Proposition 2:

High TCO to build analytics solutions Low TCO to build analytics solutions

1. High cost for development i. specialized skills needed by each time in data 1. Lower development and i. the use of wrappers to hide the complexity of
and maintenance towards mining and advanced analytics to develop, code maintenance costs with multiple languages and minimize the need for
specialized skills needed in and handle huge amounts of data diverse skills in the solution development team
collaboration and common
ii. lack of synergy and best use of the specialized ii. reuse of analytics built on any supported
advanced data analytics frameworks
skills between the teams languages across teams
iii. limited availability of functional / domain expertise iii. leverage best of breed algorithms via the
and technical skills for each development team Analytic Catalog

2. Expensive and longer i. multiple development cycles for each development 2. Minimize cost and cycle i. share best practices and lessons learnt with
customer feedback cycle team as lessons learnt and best practices are not time to incorporate Orchestration Editor
time for point solutions shared across teams ii. standardized testing process with reuse of
customer requirements
ii. limited standardization on comprehensive test input / output mapping process
cases due to to different approaches to implement with standardized,
iii. leverage common skills such as QA across
analytics collaborative analytic various domain apps from a centralized pool
iii. common skills such as QA becomes specialized in environment
a siloed analytical development environment

3. High cost and risk with the i. multiple license agreements and stack integration 3. Minimize costs and risks i. leverage enterprise license agreements and
for each analytic solution associated with the application integration (e.g.: eliminate costs
introduction of new
ii. tight coupling between data and analysis solution introduction of new associated with standalone analysis servers
technologies and such as Matlab)
iii. higher operational risk and potential failure points technologies and
programming languages ii. data extraction independent of the analytic
with the introduction of each new wave of programming languages
technology and human intervention solution
iii. encapsulate underlying technologies with
wrappers, APIs; and minimize risks with
conditional execution of analytics with
minimum human interactions

20 GE Internal
 ③ Maximize collaboration in analytics
development and operations to foster
innovation
Pain Point 3: Value Proposition 3:

Limited collaboration in analytics development and operations Maximize collaboration in analytics development and
to foster innovation operations to foster innovation
1. Inefficient time and use of i. disparate technologies and siloed environment; 1. Increase efficiency by i. one common environment to assimilate
data scientists in dealing limited collaboration and feedback loop across focusing on high value algorithms and datasets, and collaboration
with disjointed, multiple data scientists and operations across data scientists and operations
business problems
ii. lack of one standardized and common way to ii. one standard way to create analytical
analytical language
execute analytics solutions using Analytics Catalog with mix and
environment vs. focusing iii. limited ability to incorporate best of breed analytic match of analytics in different languages
on high-value business capabilities iii. reuse of best of breed analytics using
problems for the customers Analytics Catalog and ability to compare and
contrast heterogeneous analytic algorithms
2. Lack of common place to i. inability for data scientists to share and collaborate
with the algorithms 2. One common environment i. collaboration and sharing of algorithms
share analytics, algorithms ii. reuse of dataset mappings - input/output
ii. inability to share dataset mappings for use with the to share algorithms
algorithms mappings
iii. inability to share pre-processed and cleansed data iii. ability to share pre-processing and cleansing
routines for commonly used datasets

3. Inability to manage and i. lack of standardized environment and process for 3. Ability to manage and i. use of common environment for sharing
distribute secured data sets data exploration of the datasets distribute secured data sets datasets and processes for data exploration
in a centralized location ii. lack of centralized environment for sharing the ii. centralization of datasets to facilitate sharing
datasets used for analytics
in a centralized location
and reuse
iii. inability to share orchestration of analytics iii. ability to share analytic orchestrations to
provide building blocks for developing complex
analytic solutions

21 GE Internal
22 GE Internal
23 GE Internal
24 GE Internal
25 GE Internal