Complex%20Event%20Processing%20(CEP)%20WebLogic%20Time%20and%20Event%20Driven%20Product%20Family%20(TED)%20The%20Java%20Runtime%20infrastructure%20and%20Container%20for%20Event%20Driven%20Architecture%20based%20Applications

1
Complex Event Processing (CEP) WebLogic Time and
Event Driven Product Family (TED)The Java
Runtime infrastructure and Container for Event
Driven Architecture based Applications

• Wolfgang Weigend
• Principal Systems Engineer
• Team Lead Systems Engineering
• Germany, Switzerland and Austria

2
Agenda

• Product Markets and Drivers
• Product Overview
• Whats inside
• Use Case

3
The Instantly Responsive Enterprise

• WebLogic TED
• EDA (Event Driven Architecture) facilitates the
  execution of uncoupled business processes
  triggered by real-time events
• Extends SOA (Service Oriented Architecture)
  with a Sense/Respond paradigm
• Enabler for XTP (Extreme Transaction
  Processing)

4
The Instantly Responsive Enterprise
Triple play order management
Asset management
Algorithmic trading

• Large volumes of streaming events
• Need for sub-millisecond response times
• Guaranteed under heavy loads

Negative Working Capital inventory management

• Intrusion detection systems
• Military asset allocation

Reponses to calamities earthquake, flooding
5
Unique Foundational Runtime Infrastructure
(WebLogic Real Time Powered by JRockit )

• Use Case -Extreme low latency and predictability
  for Java applications Time Critical
• Benefits
• Guaranteed maximum response times
• Average response times in microseconds
• Deterministic QoS under heavy load
• No code re-writes

At first, we doubted BEA's performance claims
for WLRT 1.1. After all, if it were possible to
lose the garbage-collection latency, wouldn't Sun
have done it already? After testing WLRT 1.1 in
several scenarios, we concluded BEA's solution
works. . - Network Computing, September 2006
6
Unique Foundational Runtime Infrastructure
(WebLogic Real Time Identify, Analyze and
Resolve)

• Visualize thread transaction execution time
  lines
• Find sources of latency spikes

7
Unique Java Container (WebLogic Event Server
Exceeding EDA Expectations )

• Use Case Predictable, performant high volumes
  of streaming data Event Critical
• Benefits
• Complex event processing
• Handle high volumes of data (50k messages/sec)
• Latency optimized, lightweight Java container
• Light-weight Java deployment platform
• POJO support
• Enterprise ilities expected from an application
  server
• Scalability, Manageability, Availability,
  Security

8
Unique Java Container (WebLogic Event Server
Deterministic Container)

• Infrastructure for deterministic millisecond
  latencies under pre-determined peak workloads
• Minimize thread context switching
  andsynchronization
• Tuned I/O (e.g., socket connections)
• Real-time scheduling
• Efficient memory management

9
WebLogic Time and Event Driven Complete Product
Family

• Tuned Light Weight Java container
• Integrated CEP Engine
• Supports Business Logic as POJO
• Low latency, deterministic
• Latency metrics uniquely available
• First fully mSA-based (consumer/producer) BEA
  products

RT Development Environment
WebLogic Event Server
Applications
Event Server Infrastructure Services
QoS Latency Monitoring
Common Services
Fault Tolerance
Configuration Administration
Security
Logging
Run Time
10
Hardware and Software Platforms
11
Whats inside

• Low Latency Deterministic Container
• Lightweight Application Server
• Event Driven Programming Model
• Complex Event Processor (CEP)
• Deployment Environment
• Time and Event Driven Computing Summary

12
Low Latency Deterministic Container WLRT
Deterministic GC

• Provides QoS guarantees
• Usage -Xgcpriodeterministic Xpausetarget10ms
• Highly tuned mostly concurrent mark-n-sweep GC
• Most work done concurrently, pauses are frequent
  but very short
• Back off, split work and reschedule if nearing
  QoS limit (divide and conquer)

One GC cycle (phases 0-2, 2-4, 4-5, 5 visible)
500 ms
20 s
13
Lightweight Application Server

• Built on the BEA Core Engine
• Equinox OSGi Platform underneath
• Mix of hardened WLS subsystems and new
  technologies (Spring-OSGi, Complex Event
  Processor, etc.)
• Tuned I/O model
• Refinements to WLS network I/O model
• Blocking network I/O
• One dedicated thread/connection
• Tuned Threading model
• Minimal (zero) thread switching
• Can parallelize Rules execution or switch thread
  (if desired)

14
Event Driven Programming Model

• Applications model complex event processing
  networks
• Network of collaborating producers and consumers
  of events
• Network nodes (stages) are
• Services Adapter, Stream, Processor
• Business Object POJO (Spring Beans)
• Events POJO or Map
• Spring-based declarative assembly and
  configuration
• Applications configured in Spring configuration
  file
• Custom Spring tags for WLEvS services
• Dependency injection for additional services
• Seamless integration to legacy Spring beans

15
Event Driven Programming Model
Stream
Data Source
Adapter
Stream
Data Source
16
Complex Event Processor (CEP)

• Continuous query engine
• Processes high volumes of streaming data
• Filtering, Correlation, Causality, Aggregation
• Evaluates in Real-Time (Sub-millisecond with
  high workloads)
• SQL-based Event Processing Language (EPL)
• Extensions to handle streaming Events
• Runtime representation
• Rules compiled into Java code by Query Parser
• Actions (user code) POJOs (Spring Beans)
• Events POJOs

17
Event Processing Language (EPL)

• SQL-based
• Leverage existing skill-sets
• Easier integration to relational data
• SELECT FROM RETAIN WHERE
• GROUP BY HAVING
• MATCHING OUTPUT
• Extensions to handle streaming data
• WHEN ... THEN vs. IF ... THEN
• Streaming Windows (sliding, jumping)
• Time-based queries

18
Deployment Environment (1) Connectivity
Capabilities

• HTTP(S) for client requests/response
• JMS client for asynchronous message for client or
  data source
• Standard Java SE remote interfaces (i.e., RMI,
  JNI, JDBC)
• Adapter SDK for writing custom adapters
• Integrates with Event Sources and Sinks
• Socket framework for TCP/IP based adapters
• Easy mapping to Event Model

19
Deployment Environment (2) Comprehensive
Security Framework

• Leverages best-of-breed BEA Common Security
  Services Framework (CSS)
• Pluggable SPI model (Security Providers) for
  3rd party integration
• SSO via AUTH Provider for Cleartrust
• File based security provider for configuring
  user, passwords, and roles
• Password information encrypted
• API to retrieve user, password
• Used by http server for user authentication

20
Deployment Environment (3)Administration and
Monitoring

• Command line interface to start/stop server and
  application deployment
• Standard JMX interface for system configuration
  and monitoring
• Basic configuration status and monitoring Mbeans
  with initial release
• Tightly integrated with JR Mission Control
• Latency Analyzer

21
Deployment Environment (4)Administration and
Monitoring - Dashboard Activity Monitoring
22
Time and Event Driven Computing Summary
BEA WebLogic TED Products

• WebLogic Event Server (WLEvS)
• WLRT and WLEvS ship separately and as a bundle
• WLRT can be dropped in to other products as
  turbocharger

WebLogic Event Server
CEP, Stream, Scheduling Services
Foundation Event Services
WebLogic Real Time
JRockit DGC
Aggregate, correlate, filter data10,000 Rules /
Queries
Incoming Data Streams150,000 Messages/Seclt1K
message size

• Resulting Data fed to user Java codelt5 data
  matches filter criteria

Latency Goalslt 1 Millisecond Latency (Avg.) 30
Millisecond Latency (Max)
22
BEA Confidential
23
Use Case Financial Services Partner Fixed
Income Trader - Pricing Engine
Opportunity
Improve performance and stability of fixed income
pricing engine
Solution Powered by BEA

• Reduces Application Pause times (jitter)
• Response time improved by 50 using BEA WebLogic
  Realtime
• Sun JVM avg 16 ms, WLRT 7.5 ms
• Predictable Java runtime Performance
• Dependability improvement of 20 using BEA
  Weblogic Realtime

WLRT CE Sun Java
24
BEAs Event-driven SOA Suite Proof Points - A
Financial Services Equities Group

• Value Proposition
• Significantly Increased Throughput
• BEFORE -- 10,000 - 20,000 concurrent signal
  instances
• AFTER 50,000 concurrent signal instances
• Higher developer productivity and flexibility
• By using EPL
• Dynamic rules update through configuration

• Data sources
• Streaming market data
• JDBC data sources (MS SQL Server
• Order flow data from traders

• Output Situations of interest to clients
• Stock price moves by 3
• Spread exceeds X
• consecutive ticks in a certain direction

25
Event Driven SOA (EDSOA)In Use Case,
Architecture and product
Automated Response
Interactive Response
Real-time alerts
26
Vielen Dank für Ihre Aufmerksamkeit!