Transcript
Page 1: Open Source Enterprise Service Management (OSESM) System

Copyright 2010 Avenue Open, Inc.

Open Source Enterprise Service Management (OSESM) System

Open Source Test, Monitoring, and Management System for the Application Tier

Presentation to MIL-OSS 2010

Page 2: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 2

Agenda

• Background – Problem Statement

• Current Research Project

• Full System Vision

• Invitation to Collaborate

Page 3: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 3

Distributed Computing and ESM

• Distributed Computing - Opaque inNature; Requires Application Visibilityo Debug, Test, Monitor, Manage

• Enterprise Service Management (ESM)Systems Emerged in Responseo First Generation – Actional, AmberPoint, others

• Market Consolidation – IBM, Oracle, Progresso Second Generation – Focus on Pinpointing Specific Problems

• Transactional Analysis (BTM)• Infrastructure• Code Analysis

Page 4: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 4

ESM Market Landscape• Ever-Increasing Number of Things to

Monitor End-to-End AND Top-to-Bottomo Services, Distributed Transactions, Application

Components (Java, .NET, others)o Through ESBs, Process Engines, Supporting Infrastructureo Extend to Legacy, Military Tactical, other corner cases

• Significant Need for Cross-Layer Analytics and Root-Cause Analysis

However….

• The Gap between Customer Needsand ESM Vendor Support isWidening, not Decreasing

Page 5: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 5

Imperfect Markets• All Market Options today are COTS, Closed Source

o Limited architectural and functional extensibilityo ESM Products can Only be Extended by the Vendor

• ESM Vendors are the Bottlenecko Proprietary APIs and Instrumentation - Vendor Must Buildo Instruments Hardwired – No Multi-Purpose or Mix & Matcho Limited Support Outside of Proprietary “Vendor Stack”

• Analysis is too Narrow and Requires Duplicate Infrastructureo One Tool for Infrastructure, One for App, One for Codeo Install Different Proprietary Agents for Every System – Even Within

Proprietary Vendor Stacks!

Result – As the Need for Application Monitoring is becoming More Critical to the Success of Distributed Computing, ESM Vendors are Less Willing and Able to Meet Customer Needs

Page 6: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 6

A NEW WAY OF THINKING….

Page 7: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 7

Open Source Enterprise Service Management System (OSESM)

1. Commoditize Proven Capabilities

2. Advance the State-of-the-Art

3. Dramatically Drive Down Cost

Page 8: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 8

What We Are Researching (“R&d”) Today

Initial OSESM Framework1. Mechanisms to Collect Application Tier Data – Various Types

and Forms

2. System to Disseminate Data to the Appropriate Processing points

3. Analysis - Correlation and Root Cause Analysis System

Foundation for Comprehensive Application Runtime Test and Monitoring System

Page 9: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 9

Research Specifics

• New and varied forms of Application Tier Instrumentationo Collectors – Collect Raw Data from Application Infrastructure and

underlying Components – from Application Server, OS, and related− Project to result in the development of Agents or similar− Applicability:

– When the Mission has control of the Technology– When deep, detailed information is required

o Reflectors – Collect Raw Data from Network Devices and related− Non-invasive – Can be used when Mission does not own technology end-

to-end− Higher level data, useful for Visualization, Message Tracking, Inter-System

Analysis

• Data Dissemination Techniqueso From Instruments to appropriate Analytical Engine(s)o Foundation for Federation, needed for “Mission Thread” scale

Page 10: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 10

Advancements in Data Analysis Techniques

• Specifically Targeted to Understand Mission Thread Errors and Breakages

• Researching Sophisticated Correlation Systemo Correlate Interesting and Related Data Across All Tierso What Happened at Each Level at Any Given Point in Time

−Application−Operating System−Runtime Infrastructure−Code

• Root Cause Analysis Applied to Correlated Datao What Broke?o Where, Why, How - Specificallyo When - Proactive Notification, not post-facto Analysis

Page 11: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 11

Project Deliverables• Open, published APIs – Instrumentation and

Analytical Serverso Various forms - SOAP/WSDL/XML, REST, Binary (for Low

Resource, Low Bandwidth situations or Legacy)o Various Levels – Application Layer and Supporting

Resources

• Reference Implementations and Affiliated Documentationo Instruments – Collectors and Reflectorso Intermediary Message Processoro Analytical Servers – Root Cause and Correlation

Page 12: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 12

Benefits• Dramatically Lower Cost - Estimated TCO is 10% of today’s

COTS Solutions• Flat Out - A Better ESM

o Faster Resolution for Problems that fall between OS, Hardware, Network, Application, and Mission Layers

• Extensible - by the Communityo Instrumentation from the Community – Inexpensive, Accessibleo Build New Instruments as Needed – Info Readily Available

• Adaptable – Legacy systems, Packaged Applications, Tactical Military, Hardware Appliances, Cloud Computing, others

• Decouple Instrumentation from Analysiso Mix-and-Match best-of-breed solutions

• New or Expanded Uses – Possibilities includeo Testing & Development - Debugging Distributed Applicationso Application Security Monitoring – New or non-standard Usage Alerting,

Intruder Tracing

Page 13: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 13

Vision and Future Objective• Framework is a Foundation, a Starting Point• Full Vision – Complete System for Testing,

Monitoring, and Managing Applications at Runtime• Additional Requirements for Full System

o Metrics Processingo Discovery Mechanismso Visualization Engineo Logging Facilitieso User Portalo Administrative Console

• Ultimate Objective – Full Open Source System, with a Repository for New Contributionso Open Licensingo Open Repository for New Contributions

Page 14: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 14

Notional End-State Functional Architecture

Service EService D

Service C

Service B

Users

Service A C

C

C

C C

Persistence Store /LogsIntermediary Message Processor

UI / Portal and Reporting Engine

Network Switch R

C

R

- Collector- Reflector

Root Cause

Analysis

Correlation EngineMetrics Visualization

Page 15: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 15

Current Research and Development

Service EService D

Service C

Service B

Users

Service A C

C

C

C C

Persistence Store /LogsIntermediary Message Processor

UI / Portal and Reporting Engine

Network Switch R

C

R

- Collector- Reflector

Root Cause

Analysis

Correlation EngineMetrics Visualization

Page 16: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 16

Invitation to Collaborate

• Net Centric Computing is growing Increasingly more Complex

• OSESM Sorely Needed• We Need:

o Transition Partners – Programs or Organizations Interested in using the Results of the efforts of Current Project

o Collaboration Partners – Willing to Fund and/or Contribute to development of Full Vision System

• Interested? Please Speak to me After the Presentation

Page 17: Open Source Enterprise Service Management (OSESM) System

MIL OSS 2010 17

Wrap Up

• OSESM will Dramatically Improve the ability to Field and Manage Net Centric Applications and Mission Threadso Gather and Analyze Application Tier Data at Runtime

o Proactive Notification when Mission Thread Breaks

−What Broke, Where, How, Why

−At Every Level – Application, Operating System, Runtime Infrastructure, Code

o At Far Lower Costs – Estimated TCO 10% of COTS

• Initial Framework Under Development Today• Transition and/or Collaboration Partners Strongly

Encouraged!

Page 18: Open Source Enterprise Service Management (OSESM) System

Copyright 2010 Avenue Open, Inc.

John EmersonPresident, Avenue Open, Inc.

[email protected]

Contact Information


Recommended