Concept Expert Systems for Architectural Decisions Software Architecture

Concept: Expert Systems for Architectural Decisions Software Architecture Working Group Emerging Technologies August 17, 2004 Daniel Ellis VP Technology Dellis@go. SPS. com (703) 839 -4019

Concept Architecture ata D m e System Data Decision Support System Architecture Expert System ta a EA D EA Repository Arch it Guid ecture eline s Enterprise Architect

Potential Applications in Enterprise Architecture n n Release Planning Technology Selection Pattern Selection Budgeting

Pattern Selection Example: IBM e. Business Patterns

Example: USPTO EA Service Component-Based Architecture Task n Objective: Build Momentum for EA Program n n 70 Software Development Managers Wide range of Technical and Management Skills 100+ Systems Deliverable: Concept Application Architecture Migration Strategy n n Retire specific technology Integrate EA function with IT function

Model the Business Architecture Business Data Applications Technology

Model the Application Architecture Business Data Applications Technology

Understand the Linkages Business Architecture Data Applications Technology Application Architecture

Current Architecture is Tightly Coupled Business Architecture Business Application Data Applications Data Storage Technology Deployment

Ramifications of Coupling n Problem: Changes impact other systems n Development Concerns n n n Deployment Concerns n n Requires additional coordination Each change has a higher cost Systems deploy together Infrastructure, test, and operations are overwhelmed Very costly to back-out deployment Result: Few concurrent changes can be managed, limiting progress

Two-Pronged Migration Strategy n Assumptions n n n Targeted Investment n n n Scarce Resources Everybody wants to do a good job Enterprise Application Integration Hub Application Servers aligned with CAA Incremental Alignment n Every release of every system makes progress toward target architecture

Non-Compliant Encapsulated J 2 EE Interoperable CAA Compliant. Net Interoperable Encapsulated Non-Compliant YR 1 YR 2 YR 3

Establishing the Environment n Establishing the Target Architecture n n Policy n n Concept Applications Architecture Concept Component (or System) Architecture Enterprise Application Integration Establish custodianship Require backward compatibility Mandatory upgrade to latest available services Measure

Services Offered Services Called Exceptions Technology Changes Encapsulation and Interoperability Architecture Decision Guidance Client Server

<For each non-compliant service offered> <Services Offered> Migrate Services Offered <Service no longer used> Start [Service in Use? ] [Depend on Services not in EAI hub? ] <Candidate for compliant service> <All services in EAI Hub? > Compliant Develop CAA Service; Publish to EAI Hub [Old service no longer required? ] Retire Service <Not Ready To comply> Interoperable Publish Java Proxy to EAI Hub Backward Compatible Support both old service and compliant service Complete <All Gen Services Offered have been reviewed>

Expert System Inputs n Architecture Guidance n n n Architecture Descriptors n n n Policy Vision / Target Systems Services for the system Integration patterns Service dependencies Release Plans for other systems Compliance Scale

Concept Architecture nc nde ies ces epe i v r Se ice D ns v a r l e P S se rns a e l Services Re Patte Service Dependencies New EA Repository ms e t s y S New Patterns lans e t t Pa ase P Architecture Expert ele R System Architect Prioritized Services - Recommended Patterns - Estimated Effort - Risk Suggested Release Plan Compliance Score Polic y Rule s Com plian ce - En caps Scale u - Int erop lation e - Co mpli rability ant Enterprise Architect

Expected Benefits n Overarching n n n Consistent decision making aligned with strategy Establish a mechanism for broad-based change Reduced risk and increased accountability Supports reuse at all levels Transition n Decreased coupling Increased parallelism – more concurrent changes Reduced cost of future changes

References n Malan, Ruth and Bredemeyer, Dana; “Software Architecture: Central Concerns, Key Decisions”, Architecture Resources for Enterprise Advantage, 2002. n “An Enterprise Decision Management Architecture”, Fair Isaac Corporation, 2003. n USPTO CAA Migration Strategy, 2003. n n O’Conor, Rory; Floch, Christoph; Moynihan, Tony; Renault Tristan; Cobelles, Annie; “Prompter – A Decision Support Tool using Distributed Intelligent Agents”. IBM e. Business Patterns white paper.

Questions and Comments?

Client Server Architecture Developers Presentation Business Logic DBA Data Storage More usual than stove-pipe approach § Presentation & business logic built as a pair by same developer § Benefits of integrated DB, and simpler specification & testing § “e. Business Transformation and Integration: Reusing Assets to Drive New Business”, John Dodd and Kent Craig, CA World 2001, July 8 -12, Session Code: II 160 SN.

Client Server Architecture Coupling Developers Presentation Business Logic DBA Data Storage Applications access other applications data directly. § No guarantee that business rules are applied consistently. § Tightly coupled to foreign applications data structure. § Copyright SPS 2002. All rights reserved.