Business Object and Component Architectures: Enterprise Application Integration Encounters Complex A

1
Business Object and Component Architectures
Enterprise Application Integration Encounters
Complex Adaptive Systems

• Jeff Sutherland, Chief Technology Officer
  PatientKeeper, Inc. jeff.sutherland_at_computer.org
• http//jeffsutherland.com/

HAWAII INTERNATIONAL CONFERENCE ON SYSTEM
SCIENCES HICSS-34 January 3-6, 2001Outrigger
Wailea Resort Maui
2
Complex Adaptive Systems - cas
Holland, John H. Hidden Order How Adaptation
Builds Complexity. Addison-Wesley, 1996.

• Typical legacy system are frozen
• Chronic underinvestment in automating service
  functions
• Buyers have been insensitive to architecture,
  focused on small pieces of (customized)
  functionality.
• Fragmented, non-technical stakeholders have made
  short term buying decisions.

3
Complex Adaptive Systems

• Composed of interacting agents which respond to
  stimuli.
• Stimulus-response behavior can be defined in
  terms of rules.
• Agents adapt by changing their rules as
  experience accumulates.
• Agents can be aggregated into meta-agents whose
  behavior may be emergent.

4
Enterprise Systems are cas

• Business entities are examples of complex
  adaptive systems.
• Modification time is on the order of months or
  years, roughly time required to change software.
• Automating business processes renders parts of
  the business in software.
• Business systems have severely constrained rule
  sets, making them an ideal test bed for cas
  concepts. 

5
Powerful EAI strategy is capturing software

• It is often taken for granted that as systems
  evolve over time they tend to become more
  complex...
• Single systems may grow by increases in
  structural sophistication the system steadily
  cumulates increasing numbers of subsystems or
  subfunctions or subparts to break through
  performance limitations, or to enhance its range
  of operation, or to handle exceptional
  circumstances.
• Or, it may suddenly increase by "capturing
  software" the system captures simpler elements
  and learns to "program" these as "software" to be
  used to its own ends... (Arthur 1994)

6
Hollands Key Concepts and Business Object
Architectures

• Aggregation (property)
• Tagging (mechanism)
• Nonlinearity (property)
• Flows (property)
• Diversity (property)
• Internal models (mechanism)
• Building blocks (mechanism)

7
Aggregation

• Aggregation - the basic mechanism in object
  modeling
• Forming components out of objects is aggregation.
  
• More important are emergent properties of
  intelligent objects
• Meta-agents (an enterprise)
• aggregates of agents (enterprise systems)
• emergent behaviors (revenue, profitability, and
  cash flow, the indices of value creation).

8
Tagging

• Facilitates the forming of aggregates XML, CORBA,
  COM, EJB
• Facilitates selective mating, messaging.
  Firewalls use tagging to preserve boundaries
  between aggregates.
• Object identity used to componentize object
  models. Enables filtering, specialization, and
  cooperation.
• Develop hierarchical aggregates that exhibit
  emergent behaviors (like an operating system).

9
Nonlinearity

• Catastrophic and chaotic behaviors like traffic
  flow on the Internet.
• Brownouts, system loadings, scalability effects
  are often nonlinear.
• Predator/prey interactions
• Arrival, proliferation, and destruction of
  viruses on the Internet
• Revenue prediction in an enterprise financial
  system
• Rate of construction of software

10
Flows

• Workflows
• Tags condition flows
• Flows typically have a multiplier effect.
• money injected into the economy
• email or other message flows on a network.
• Recycling effect
• individual pieces evolve, die, replaced, reused
• living software is constantly changing due to
  flows, as rivers change their course.
• dead software is eventually detritus that is
  expelled from the enterprise organism.

11
Diversity

• Persistence of an individual agent depends on the
  ecosystem of agents that surround it.
• Evolution causes convergence of system
  architectures.
• Emergent patterns that reappear again and again
  in widely disparate environments.
• Usefulness in business object systems arises from
  interactions between diverse agents as in human
  societies.

12
Internal Models

• Utility of cas is enhanced if the system can
  learn from experience and adapt behavior.
• Systems must develop and act on internal models
  that simplify the external world.
• Systems infer results of actions before they are
  taken to choose actions that have productive
  results.
• Prospects for longevity of software systems
  depend on this capability, just as in living
  systems.

13
Building Blocks

• Reuse is dependent on building blocks.
• Basis of Moore's law in hardware production.
• Could be the basis of dramatic improvements in
  software productivity.
• Building blocks are the basis for generation of
  internal models and are essential to the
  construction of adaptive enterprise systems.

14
EAI Case 1 Workflow engine, intelligent
adapters, and XML messaging

• New technology was superimposed on the old in
  order to enable
• (1) interaction between web HTML clients and
  mobile WML clients,
• (2) utilization of standard transform mechanisms
  with XSL/XSLT,
• (3) easy integration with future systems in
  electronic marketplaces through XML, and
• (4) validation and language mapping capabilities
  available with XML DTD and XML Schema tools.

15
EAI Case 1Workflow drives backend

• Business logic is encapsulated in a workflow
  which drives backend Unisys LINC system.
• Intelligent adapters provide for the glue that
  links the external applications to the workflow.
• Adapters transform XML message formats into other
  data formats or into objects and are able to take
  different actions based on the content of the
  message.

16
EAI Case 1Workflow drives frontend

• Conductor workflow engine sends XML messages to a
  frontoffice adapter which insulates the
  frontoffice Sun Forte 4GL system from the
  workflow layer

17
EAI Case 1 and cas concepts

• Two legacy systems are aggregated into a single
  system by adapters controlled by the workflow
  engine.
• Nonlinear behavior induced by actions of goal
  seeking agents is avoided in this system by
  essentially hardcoding workflows and tuning them
  prior to production.
• Flows are automated by the workflow engine and
  routed using tags supported by the infrastructure
  of XML. This is the key innovation of this
  strategy. Workflow captures legacy systems.
• Diversity is managed by shielding the workflow
  engine with robotic clients.
• Internal models are essentially hardcoded,
  preventing dynamic adaptation.
• Building blocks used in aggregation are enabled
  by workflows interacting with adaptors that
  manipulate legacy systems.

18
EAI Case 2 Mobilizing Healthcare Information

• Case/control studies show that minimal healthcare
  automation can triple compliance, cut cost in
  half, and improve clinical outcomes by 50-100.
• Physicians are inherently mobile and highly
  resistant to PC based automation. Mobile devices
  at the point of care will be required.
• Radical automation via intelligent agents could
  create an effect similar to improvements in the
  auto industry. Just as cars now run routinely to
  100,000 miles, the average person will cruise to
  100 years of age with a high quality of
  performance. Lower health care costs per unit of
  value will be a byproduct.

19
Primary Care Physician Computerization Levels
Renata Bushko. AHCPR Data, FHTI 1997.
20
Clinical OutcomesTop 10 Causes of U.S. Deaths
1994
Medication Error Annual Cost 76B Bootman, Lyle.
Arch Int Med 115181949-1956, 1995.

• Medication deaths might be reduced from 106,000
  to less than 25,000 with automation.
• Deaths from meds are significantly underreported
  so effect may be more dramatic than expected.

Lazeron J, Pomeranz B. Corey P. Incidence of
Adverse Drug Reactions in Hospitalized Patients.
JAMA 2791200-1205, 1978.
21
EAI Case 2 PatientKeeper Platform
Handheld

• Patient centric UUI
• Integrated applications 1..n
• Device independent XML datastore
• Application API
• Synchronization to clinical repository

Patient Manager
Alert Manager
WebApp
Inter Application Framework
Data Store

• Web-based repository
• Web-based preferences
• Web-based central administration

Transport Protocol
Synchronization Protocol
GlobalSync

• Synchronization to clinical repository
• Java/XML workflow engine with client APIs
• Personalization engine to forward user-specific
  data to mobile device
• Mapping template APIs to convert client datainto
  standard Virtmed format
• Client adaptors with XML parsing functions for
  communication between foreign clinical
  repositories and Virtmed platform
• Clinical repository with application objects
• Patient management system
• Proven EMPI and Interface Engine
• Full-featured proven clinical database

Transport Protocol
Cerner Transport Protocol
CHCS Adapter
Other Adapter
Clinical Repository
Webtop Protocol
Cerner Transport Protocol
LegacyApps
LegacyApp
LegacyApp
Component API
Virtmed Workspace
Patient Management
Interfaces
EMPI
Clinical Repository (Oracle)
22
EAI Case 2 and cas concepts

• The architecture aggregates multiple
  heterogeneous systems through component objects,
  adapters, or messaging interfaces.
• Flows are handled by workflow engines at both the
  synchronization server and clinical repository
  layers of the architecture.
• Tagging is supported by XML infrastructure.
• Diversity is shielded from the mobile
  infrastructure by the clinical repository.
• Building blocks are aggregated with a wide
  variety of integration patterns and mechanisms
  for interoperability. The system is complex
  enough to induce nonlinear behavior but this must
  be managed by manual tuning or coding.
• Internal models are implicitly defined by
  hardwiring components for specific applications.
  Mobile devices used as remote controls enable
  adaptive behavior. Key innovation is that user
  can dynamically modify workflow.

23
EAI Case 3 Big Workflow

• Big Workflow initially designed to cross
  multiple applications and multiple vendors to
  support patient flow across an healthcare
  Integrated Delivery Network (IDN). Similar
  systems have been implemented or are under
  construction in manufacturing and other vertical
  application domains.
• Workflow must be managed across across hundreds
  of large applications provided by dozens of
  vendors.
• Centralized management of business processes are
  imposed on all vendor applications.
• Business processes can be modified globally
  without modifying vendor systems.

24
Big Workflow is Federated, Heterogeneous,
Distributed, Enterprise Workflow

• Any server on the Internet can host any component
• Process manager, workflow engine, observers,
  worklists
• W3C standards (HTTP, XML, SOAP everywhere)
• No single point of failure
• Any system that supports an XML remote procedure
  call (SOAP) can participate

At the OOPSLA97 Business Object Workshop,
Santanu Paul presented Essential Requirements for
a Workflow Standard. He reviewed work on Rainman
A Workflow System for the Internet at IBM T.J.
Watson Research Center and pointed out
deficiencies in the WfMC architecture.
25
Workflowand Agents
Browser/Mobile/Wireless Devices
XML
HTML
HTML
Transport
UI Metadata
View Layer
Metadata
XML
XML
Application Model Layer
Business Logic
Workflow Engine
XML
Patient Object
Domain Layer
Persistence Layer
Data Access
26
Enterprise Workflow
Process Templates
1. Treat a patient
Process Manager
2. Who is responsible?
3. Where is their worklist?
Policy Manager
LDAP
4. What happens next?
5. Insert work item.
Worklist Proxy
Workflow Engine
Worklist
6. Do it!
27
Create Work
1 Create work (XML context)
ProcessManager Servlet
WfLDAP
2 Init LDAP entry
http
3 Invoke Process ManagerCreateWork (context,
LDAP ref)
ProcessManagerImpl
Persistent Worklist Items
4 Invoke WfEngineGetNextState (context)
WfEngine
WorkListImpl
5 If active work, put on worklist
7 Insert work (return ID for WfNewWork)
6 Insert(context, WorkItem)
WorkListProxy
WorkList Servlet
http
Note If Owner is not Performer, must create
Observer entry in Owner WorkList and call
Performer WorkList to Subscribe
28
Big Workflow lays foundation for Agent driven
integration

• Autonomous controls own behavior
• Interactive talks to other agents, humans,
  machines
• Adaptive, reactive senses and responds to
  environment
• Mobile transports itself across platforms
• Proxy - may act on behalf of someone or something
• Goal oriented pro-active and purposeful
• Rational - able to reason or understand
• Intelligent - state is formalized (i.e., beliefs,
  goals, plans, assumptions)
• Temporally continuous - is a continuously running
  process
• Character traits believable personality/emotiona
  l state
• Transparent and accountable
• Learning and evolving experience changes
  behavior
• Cooperative - e.g., can coordinate, collaborate,
  and negotiate
• Rugged - able to deal with errors and incomplete
  data
• Trustworthy - adheres to Laws of Robotics and is
  truthful

OMG Agent Working Group. Agent Technology Green
Paper. OMG Document ec/99-08-06, Version 0.8, 20
August 1999.
29
EAI Futures Objects and Agents
Jim Odell, Chair, OMG Agent Working Group
James Odell. Objects and Agents How do they
differ? Distributed Computing, 1999 (in press).
30
EAI Futures Collaborative Agents
31
EAI Futures Collaborative Agents

• Business object (BO) components use a
  multilayered architecture to support agent
  technology.
• The first layer contains knowledge about goals to
  seek.
• The second layer advertises services to other
  business object components that may be performed
  by this BO and are associated with the goals of
  the BO.
• The third layer provides the operations used to
  perform the service and these operations may be
  packaged into workflows.
• The fourth layer contains references to
  operations performed by other BOs.
• Coordination between BOs is carried out by
  conversations based on well-known coordination
  strategies

32
Summary

• New technology will provide automated, adaptive
  behavior to the enterprise by capturing
  software in legacy systems.
• Legacy system services will be widely embedded in
  global internet systems supporting supply chain
  integration just as mitochondria are in humans.
• Workflow tools will orchestrate large scale
  system behaviors and serve as reptilian brains.
• Collaborative agents with goal-seeking objectives
  will provide a neocortex and use reptilian brains
  to execute basic behaviors.