Case Study: Semantic Integration as the Key Enabler of Interoperability and Modular

Architecture for Smart Grid at Long Island Power Authority (LIPA)

Predrag Vujovic, Stipe Fustar, Phillip Jones, Fran Clark

Overview • Real-life Case Study of LIPA Architecture and

integration approach in their Smart Grid Program

• Our Story: – LIPA Smart Grid Architecture approach – Semantic Integration as the Key Enabler of

Modular / pluggable architecture – A real-life solution change scenario (switching

from one solution provider to another one / (“unplug” one and “plug” another system)

LIPA Model-driven Architecture Business Drivers

• Establish a loosely-coupled SOA architecture through : – Leveraging an Exchange Model (EXM) for model-driven

“development”, that … • mediates all interfaces through a LIPA Enterprise Semantic Model (ESM), … • which is based on available industry standards (e.g. CIM)

• Use of a model-driven design and development process to: – Speed development process – Improve reusability – Improve governance and change management

• Require that any new vendor applications: – Conform to LIPA end points (canonical interfaces, staging areas) as

much as possible / practical – When not possible, conform to applicable industry standard interface – Publish interfaces / APIs so that knowledge of underlying database

structures is not required for integration (transactional or analytical)

Key Elements of LIPA Semantic Integration • Centrally Managed Semantic (Data) Model

– Heterogeneous interfaces mediated through common model – Based on industry standards (IEC CIM is the key reference model)

• Centrally Managed Semantic Mapping and Business Rules – Integrate & Reuse Business Rules, transformations, mappings – Automate gap analysis, documentation

• Centrally Managed Development and Run-Time Deployment – Generate ready-to-go SOA services – Continuous testing – Deploy into any runtime environment – Automate impact analysis on change

6

LIPA Model-driven Architecture Benefits

• Reduce cost of implementations and integrations – including maintenance / change management

• Reduce risk to implementations and integrations • Increase speed of implementations and integrations • Improve ability to solve business problems by choosing best of breed

applications and services • Avoid vendor- and technology lock-in’s • Support Multiple Service Providers • Architecture:

– “Near Plug and Play”, Flexibility, Agility & Portability – Consistent semantics for data in-flight and persisted data stores – Event-driven – Flexibility of Business Intelligence Options – Open to new technology, solutions, applications

Foundation – Model-driven process

• LIPA has adopted a model-driven process for defining, designing, developing and deploying: – Services on the Enterprise Service Bus – Persistent data stores for analytics (ODS, Data Warehouse, Datamart)

• The Model-driven approach leverages industry standards (e.g. EC CIM) wherever possible to: – Promote reusability – Accelerate development cycles – Facilitate visibility, governance and change management

• Four key models – Enterprise Semantic Model – Service Model – Exchange Model – Data Model

• Process & Governance

Slide 10

Model-Driven Information, Integration & Intelligence

Semantic Modeling

Semantic Mapping Enterprise Integration View

Focus is on Semantics

Meta-data

Enterprise Model

Data Services

Data Services

Relevant Standards

(e.g. IEC CIM MultiSpeak)

Sing

le /

Co

mm

on

Ente

rpris

e Se

man

tic

Mod

el

LIPA

Spe

cific

U

tility

Dom

ain

Dict

iona

ry

Interface Tier

Data Tier

Business Logic Tier

Process Integration (Simple & Complex Data Exchanges, Transacting

processes)

Business Intelligence

(e.g. Spotfire)

Model Driven Development Integration Services & Data-store Design Design – Develop – Deploy Cycle (information perspective)

Design- & Development

Semantic Model Application

Meta-Data

etc

CopyBook

XSD

API

LIPA ESM

Reference Model

Reference Model

Reference Model

Reference Model

Reference Base Model

Reference Base Model

Reference Base Model

Reference Base Model

Trfrm

Trfrm

Trfrm

Trfrm

Business Vernacular

Gap analysis & extension

Gap analysis Extension

Gap analysis Extension

Run-Time

SM Context Service

Design Inputs

Test Map once & Develop

Gap Analysis

Impact Analysis

Mapping Reports Deploy

Model Driven Development

Versioning

Exchange Model

Process Design Inputs

import

Impact Analysis

DXSI Run-time Components

Industry Standards

ODS, ODW

import

DB Schema

Implementation Model (ODS, ODW)

LIPA EDM Workflow : “Lossless” Metadata Common Data Model

Integration Team Operations Team Local requirements

The Information Architecture team customizes the common data model with local requirements

The Information Architecture team implements the exchange model integrating with the necessary systems and performs initial data transformation testing

Components for production are generated by the integration team, re-tested and passed to the operations team for deployment

Information Architecture

The Integration team performs complete testing, including process execution and exception handling.

XMI

Operations Team deploys Components in the production environment.

ESB

DXSI

© Xtensible Solutions Inc. Continuous Testing

Enterprise Semantic

Model

Stateless for scalability

Runs in any Java container

Connects to any bus

Distributed run-time Testing

Centralized Mapping, Design

Centralized Data Model

Enterprise Exchange

Model

Schema Schema Schema

LIPA ESB Platform by Business Domain

LIPA SP Integration-Ready Gateway

• Energy Trading – LIPA and three service providers – Repeatable patterns and solutions – Staging areas

• “Privacy” of participant’s internal data and systems

• Parallel development • Ready for future “plug and play”

– Centrally managed data model – Plans to introduce event-driven

patterns

• T & D – OMS, MDM, ODS / ODW, Customer

Communication projects data exchanges

– Centrally managed data model – Repeatable patterns and solutions – Event-driven patterns – Use of Semantic Integrator DXSI and

Sonic ESB – Ready for future “plug and play” – Proxy/Reverse proxy

• Secures access to façades of canonical and adaptor-enabled service

• Host SP queues, if applicable

Key Integration Patterns

15

Key Integration Patterns - Pluggable Endpoints

16

Key Integration Patterns – Integration Ready infrastructure

17

Adaptor / Canonical Pattern (P)

Canonical Service Patterns

(P) (C)

Adaptor Service Patterns

(P) (C)

Adaptor / Canonical Pattern (C)

Integration Ready – Plug-n-play

Key Integration Patterns – Library of Endpoints

18

Adaptor / Canonical Pattern (P)

Adaptor / Canonical Pattern (C)

Library of integration-ready / pluggable endpoints and services increases by each project

Canonical Service Patterns

(P) (C)

Adaptor Service Patterns

(P) (C)

Solution Change Scenario: From Vendor A to Vendor B

19 Unplug OMS Solution from Vendor A

Solution Change Scenario: From Vendor A to Vendor B

20 Plug OMS Solution from Vendor B

Endpoints unaffected by OMS solution change

Key Take-Away Points • Innovative Integration approach with benefits of

– “Near Plug and Play” for systems and Analysis Solutions – Model Driven Development, End-to-End – Benefits of automation for integration, testing, maintenance, updates – Significantly Lower Life Cycle Cost and more effective system

deployments • Model-driven approach that leverages Industry Standards (CIM) for

interoperability • Scalable (Structured, planned, model-driven approach) • Semantic understanding is guaranteed (explicit, not implicit) ;

– availability of strongly typed syntactical interfaces is not a requirement for success any more

• Easier updating and tracking of standards development • Modular / pluggable architecture has been proven in a real-life

change scenario

Thank You • Predrag Vujovic

– pvujovic@lipower.org • Stipe Fustar

– sfustar@powergrid360.com • Phillip Jones

– pjones@xtensible.net • Fran Clark, Arpeggio Technology

– franclark@comcast.net