tibco brforum09 whats different about rules in cep

8/3/2019 Tibco Brforum09 Whats Different About Rules in Cep

1/25

Business Rulesand Business Events

- where CEP Helps Decisions

2009 TIBCO Software Inc. All Rights Reserved. Confidential and Proprietary.


2/25


3/25

Introduction

Speaker: Paul Vincent, TIBCO Software CEP GroupCo-author and contributor to rule standards (PRR, RIF... DMN)

Co-chair EPTS RAWG

TIBCO blogger via http://tibcoblogs.com/cep

Motivation: identify event processing (/EDA) as a


3 2006 TIBCO Software Inc. All Rights Reserved. Confidential and Proprietary.3

Useful augmentation / alternative to the defaultBPM / SOA / decision service regime

A closer mapping to the business rules mantra

Set of useful business benefits (BAM, OI, ...) too!

Disclaimer/caveat: CEP technologies workalongside other (BPM, SOA, BRE) technologies...


4/25

Introduction - TIBCO

Events

Data

Messaging

BPM & SOAEDA

TIBCO EMSTIBCO RV

TIBCO ActiveSpaces

TIBCO ActiveMatrixTIBCO iProcess

TIBCO BusinessEvents

Real time / low latency Data time / high volume



Rule Authoring

Processing

Visual Analysis

EventDashboards

Predictive Analytics

TIBCO BE Decision Manager

TIBCO BE Syndera TIBCO Spotfire

TIBCO Spotfire S+

Real-time Analytics

Real-time Patterns

Real-time Decisions


5/25

Introduction - CEP

Saved my 401K by identifyingan economic pattern via

the metric of countingrailway trucks



"Future of business rules is CEP"

" DM platform needs data preparationand decision refinement

and also state ie combining with CEP...

Why not bring them together inactive on / always on rule processing? "


6/25

Agenda

1. Business Rules (per BR Concepts, 3 rd Ed)

2. Events vs Business Rules

3. Simple Event Processing

4. Complex Event Processing



5. rc ec ure vs us ness u es


7/25

Business Rules 101

Business Rules can be:

Restrictions

GuidelinesComputations

e.g. X must have Y

= constraints on businesse.g. X should have Y

= suggestions forbusinesse.g. X = f(Y)



InferencesTimings

Triggers

,. .

= business knowledgee.g. do X at time T

= temporal dependenciese.g. when X occurs do Y

= event dependenciesCan be combined:when E occurs within time T,if X infer that Y must have Z

with value f(Z) ...


8/25

Business Rules 102

Business Rules can be:

Restrictions

GuidelinesComputations

Business rules are definedthrough a business

ontology: terms and facts

Some facts may be events(state changes,



InferencesTimings

Triggers

observations, incidents)e.g.Flight arrivesAt Airport

Rules are enforced asevents occur

We want to predict whenrules will get broken


9/25

Events are key tobusiness rule enforcement / evaluation

Business rules drive process definitions

Business rules drive decisions made in business

processesMapping from business rules to processes anddecisions is easier from an event perspective



e.g. An Order over $1000 must not be accepted on credit without a credit check

New Order event Change Order event Change Currency /exchange rate event

Change Rule

Condition event

Rule applies to

order process...

Change Limit (Rule

Parameter) event


10/25

Events in business rule enforcement can be simple...

New Order event

BPM: order process

Rule Change event

BRMS: rule management process

Tight process-decisionservice development relationship



e.g. An Order over $1000 must not be accepted on credit without a credit check

Decision Service: process rules toenforce business rule

BPM: sequential activitiesespecially useful for manualprocessing...

Tight process-decisionservice coupling


11/25

Events in business rule enforcement can also be complex

New fraud event(aggregated fromother events)

BPM: order process



e.g. An Order over $100 and with a fraud possibility > 45% must not be accepted without a customer identity check

Decision Service: process rules toenforce business rule


12/25

But... Processes are just aggregating events too!


EP: order state

New Order event



e.g. Order (new? changed?) value $X and fraud possibility Y%

can only be accepted with a customer identity check

Event processing rules enforcebusiness rules / decisions


13/25

But... Processes are just aggregating events (3)!


EP: order states: defined, accepted, fulfilled, changed, ...

New Order event



e.g. Order (new? changed?) value $X and fraud possibility Y%

can only be accepted with a customer identity check

Event processing rules enforcebusiness rules / decisions+ progress order stateState change in

order


14/25

But... Processes are just aggregating events (4)!


EP: order state

New Order event

State set to suspended



e.g. An Order over $100 and with a fraud possibility > 45% without a customer identity check that has already been accepted then shipments must be suspended until the fraud possibility is reduced to below 5%

Event processing rules enforcebusiness rules / decisions+ advance order state+ exceptions

New eventchanges fraud to75% after Order

accepted...


15/25

Why Use a CEP Platform for business rule automation?

e

Business Event

warnings precede threats

CEP providesquicker response to detected issues



P o

t e n t

i a l B u s

i n e s s

V a l u

Resulting Complex Event Measured

Root Cause / Correlation

Corrective Decision Made

Action Taken

Action time


16/25

CEP in Action: Investment Banking

Algorithmic Trading CEP Application

State Engine

InferenceRule Engine Rulebase

State Model



Front Office Middle Office Back Office

Data Feeds

(events)

Archival etc

DatabaseSOA

Services

Compliance CEP Application

State Engine

InferenceRule Engine Rulebase

State Model


17/25

CEP

Rulebase State Model

Event Cache Other Data

New

Orderchannelled

CEP Processing activity is automatic

Update& TrackStatesClassification

& filter(Re)Assess Order

vs other orders

Reg Compliancecheck required?



4. Comparison rules run againstorder by rule engine, newinformation inferred

5. State transition rules executedto check for any state changedue to new inferred data.

6. If order can be completed, runSEC compliance rules

17

1. Order event arrives over JMS / EMS or RV on a particularchannel

2. Order event classified intoappropriate class, stored incache for future reference

3. Order state assessed andupdated due to event details


18/25

Decisions are Event Driven

Complex events require monitoring the simpleevents and decisions

Identify patterns

In real-timeThe decisions share events, data

Credit Card application patterns Consumer Fraud possibility



All about Situation Awareness: business monitoring and insightSo why keep the data in a separate system?

Traditional answer: too complex for IT to design and managedecisions and data models in the same system

Current solution: CEP technologies combine rules, events andreal-time data stores to maximize performance


19/25

Example Architecture for Real-time Decisions

CEP Rule Agent

State Engine

Inference

State ModelCEP Query Agent

DecisionManagement

CEP Editors Managed Entities

RealtimeDashboard

AJAX

Business Events drivebusiness rule execution

Rule maintenance &execution using BRE, BRMS

features

Real-time analytics using real-time rule monitoring

+ statistical functions / agentsdriving rule & score changes



Distributed CEP Framework

E v e n

t S o u r c e s

Event Bus

Rule Engine

Rulebase

Data Grid E v e n t

S i n k s

Query Engine

Query List

Event Model Class Model Libraries

Champion Challenger thrumultiple ruleset control / multiple decision agents

on-t e- y

Operational event storeprovides event warehouse for

real-time historic patterndetection

IT-friendly model-drivenengineering via easy-to-understand state, query,

rules, concepts,


20/25

SOA vs EDA patterns for Decision Services

Business Rule

Service Invoker Process

Business Event

Event-drivenBusiness Rule

/ Decision Engine

Business Event



ec s on ng ne

Decision Result

Decision Result

time


21/25

SOA vs EDA patterns for Decision Services

Define Business Event Define Business Event

Define Decision Event

Define associated RuleDefine App Server /web service

Define Service / Ruleset



Define Decision Result

Designand

developmenttime / cost

orchestration

Define associated Rule


22/25

Event and data patterns for Decision Services

Service Invoker Process

Business Event


/ Decision Engine

Business Event

Data Retrieve Process Data Grid Access



Business Rule/ Decision Engine

Decision Result

Decision Result

runtime

Data Store Process


23/25

Decision Logic patterns for Decision Services

Pre-defined and ordered

Decision Service invocations

Business Event


/ Decision Engine

Business Event

Data Retrieve Process Data Grid Access



Business Rule/ Decision Service 1

Decision Result

Decision Result

runtime

Data Store Process

Business Rule

/ Decision Service 2


24/25

High Scaleability patterns for Decision Services

Decision Service invoker

Business Event Business Event

1

Data Retrieve Process

1 Data Grid Access


/ Decision Agent


/ Decision Agent

2

2




Decision Results

Decision Results

runtime



25/25

CEP and Decisions Summary

Business rules are used in CEP applications covering

Sense & Respond

Track & TraceSituation Awareness

Users should model events independent of infrastructure



Flexibility / EDA vs Commonality / SOA

" We need to move from beingprocess centric

to moreinformation centric"

tibco brforum09 whats different about rules in cep

Documents