Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Next Generation Grid Enabled SOA:

670 views

Published on

  • Be the first to comment

  • Be the first to like this

Next Generation Grid Enabled SOA:

  1. 1. <Insert Picture Here> Next Generation Grid Enabled SOA: Delivering Continuous Availability and Linear Scalability for Applications in a Service Oriented Architecture Dave Chappell VP & Chief Technologist, SOA
  2. 2. About the Speaker 2
  3. 3. Agenda/Outline • SOA Today: Drivers for Change • SOA Tomorrow: SOA Grid • State Management in the SOA Grid • SOA Grid – Advanced Capabilities • QoS and Distributed SOA Processing • Relocatable Stateful Orchestrations (BPEL) • Not Your MOM’s Bus • New Model for Scaling SOA • Summary 3
  4. 4. SOA Today: Level Setting What were we suppose to get from SOA? • IT Management Paradigm Shift • Reduce Cost and Complexity Large XML Payloads • Service Enablement of IT Assets Unexpected Usage • Leverage investment, reuse Demands • Business Agility Meeting SLA • Better align with the Business Needs Expectations • Automate Business Function • Business Process Orchestration Tearing Down / De-coupling • Composite Applications Silos • Flexibility • Loose Coupling, Modularity Sharing Information • Easily integrated, upgraded, replaced Across Multiple Services • Re-Focus on Innovation, New Business Services 4
  5. 5. Drivers for Change: The Evolving Problem Set 50,000 foot view • SOA Architect: Tearing Down Silos • Reuse of Shared Services • New Flexible Business Processes • IT Operations: Deployment Complexity • How many configurations of servers? • When and how to add more flavors? • Cost and Efficiency • Datacenter resource utilization • Usage typically 13% – 17% • Virtualization only solves part of the problem Tight Coupling & Contention Between SOA Architect & IT Operations 5
  6. 6. Continuing Challenges • Predictable Scalability • Continuous Availability • Infrastructure AND Services • Reliable QoS • Conventional Reliable Messaging only solves part of the problem • Availability / Reliability of transient application data • Shared context across load balanced / HA services 6
  7. 7. Complexity of Managing “state” in SOA Applications Hard-wired Shared Complexity Context DB Tight Coupling i ty ex State passing via pl XML Payloads C om e ic Cookies + rv Servlet Se Session Stateless Service Sophistication, Stateful Longevity 7
  8. 8. Agenda/Outline • SOA Today: Drivers for Change • SOA Tomorrow: SOA Grid • State Management in the SOA Grid • SOA Grid – Advanced Capabilities • QoS and Distributed SOA Processing • Relocatable Stateful Orchestrations (BPEL) • Not Your MOM’s Bus • New Model for Scaling SOA • Summary 8
  9. 9. The SOA Grid • State-aware continuous availability for service infrastructure, application data, and processing logic • Predictable scalability • Scales out linearly, whether 2 or 2,000 servers • Heterogeneous Environment • High-end / low-cost commodity hardware • Dramatic overall increase in performance and throughput • Linearly scalable shared memory and logic • Reduced dependency on disk persistence • Reduced operations cost • Reduced deployment complexity • Less flavors of servers to deploy 9
  10. 10. SOA Grid - Primary/Backup synchronization Non-storage-aware Storage-aware Datagrid clients Datagrid servers Primary Update/Fetch Node P Application Ack/Nack Backup Node B Application Application 10
  11. 11. SOA Grid - Primary/Backup synchronization Non-storage-aware Storage-aware Datagrid clients Datagrid servers Application Ack/Nack Update/Fetch X Backup P Node Application Primary B Node Application 11
  12. 12. Asynchronous DB Updates Non-storage-aware Storage-aware Datagrid clients Datagrid servers Primary Node Update/Fetch P Ack/Nack Application Backup Node B Application DB Grid Application = Write Behind Queue 12
  13. 13. Near-Cache Optimization Non-storage-aware Storage-aware Datagrid clients Datagrid servers Primary Node Fetch P Application Cache Backup Node B Application DB Grid Application 13
  14. 14. SOA Grid – Advanced Capabilities • Co-locate service code with grid data • Load balance and dispatch requests appropriately • Availability and failover of Stateful services • State Passing Model Redefined • BPEL dehydration into the grid • Relocatable BPEL processes 15
  15. 15. SOA Grid Primary Update/Fetch Node P Ack/Nack Service Backup Node B ESB Service Service SOA Grid BPEL Process 16
  16. 16. Stateful Service load balancing Service ‘A’ Invocation Primary Update/Fetch Node P Ack/Nack Service A Backup Node B ESB Service A1 Load Balanced SOA Grid BPEL Process Services 17
  17. 17. Stateful Service load balancing Service ‘A1’ Invocation Primary Node P Service Backup A Node Fetch B ESB Service A1 Load Balanced SOA Grid BPEL Process Services 18
  18. 18. Stateful Service availability/failover Service ‘A’ Invocation Primary Update/Fetch Node P Ack/Nack Service A Backup Node B ESB Service A1 Load Balanced SOA Grid BPEL Process Services 19
  19. 19. Stateful Service availability/failover Service ‘A1’ Invocation X Primary Node P Service Backup A Node Fetch B ESB Service A1 Load Balanced SOA Grid BPEL Process Services 20
  20. 20. BPEL Dehydration Example BPEL Server 1 Process 1 Data Flow 1 Invoke 1) BPEL Server 1 Process 1 invokes Service 1 setting callback URL to LBR ESB 3 2) BPEL Server 1 Process 1 dehydrates to Load Balancer Service 1 Grid Callback 3) Service 1 Invokes LBR 4) LBR invokes BPEL Server 2 Process 2 BPEL Server 2 5) Server 2 Process 1 rehydrates process Process 1 data from grid and continues 4 ESB Service 1 and 2 can store their own Service 2 data in the SOA Grid but they do not require access to the BPEL dehydration store in this example. 5 P B SOA Grid 21
  21. 21. Relocatable BPEL processes 1 234 Dehydrate •Activate/Rehydrate BPEL Svc 5 process where the primary P data resides •Lightweight signaling across nodes Lightweight signal B •State transitions using grid triggers •Tradeoffs: BPEL payload B rehydration/serialization vs. P biz object serialization 1 234 Svc Rehydrate 5 SOA Grid 22
  22. 22. Agenda/Outline • SOA Today: Drivers for Change • SOA Tomorrow: SOA Grid • State Management in the SOA Grid • SOA Grid – Advanced Capabilities • QoS and Distributed SOA Processing • Relocatable Stateful Orchestrations (BPEL) • Not Your MOM’s Bus • New Model for Scaling SOA • Summary 23
  23. 23. SOA Grid – Not Your MOM’s Bus Conventional Messaging for QoS Web Service WS-A ddr Inbound <ReplyTo> Callback Portal CRM ERP 1 2 3 4 JMS / MOM / WS-RM Core 5 BPEL CEP BAM Rules Web Service Provider 24
  24. 24. QoS and Distributed SOA Processing Not Your MOM’s Bus • Rethink Assumptions of Conventional MOM • JMS, MQ, RV, WS-RM, etc • Reliably pass data from one place to the other • If its already in the grid, and all interested parties can access it, then why “send” it? • For large data sets, why create JMS or SOAP messages simply to put things into the bus and take them out again? • Rather make changes in the grid, and notify interested parties of the change 25
  25. 25. SOA Grid – Not Your MOM’s Bus Why send it when its already there? Web WS-Addr WS-Addr Service <ReplyTo> <ReplyTo> Consumer Callback Callback Portal CRM ERP 1 2 3 4 5 BPEL CEP BAM Rules Web Service Provider 26
  26. 26. That Being Said… Still Plenty of Use Cases for Conventional Messaging WS-Addr WS-Addr Web <ReplyTo> <ReplyTo> Service Callback Callback Portal Consumer CRM ERP JMS / WS-RM / EDA 1 2 3 4 5 JMS / WS-RM / EDA BPEL CEP BAM Rules Web Service Provider 27
  27. 27. Rule of Thumb • Still need MOM for – • Familiar client API / usage model • Ordering • Pub/Sub • Avoid putting state in Queues where it doesn’t belong • Avoid “sending” stuff when it doesn’t really have to travel anywhere • Subscribe to state changes in the grid using observer pattern 28
  28. 28. Federated SOA Grid Autonomy of Control, Consolidated Business Web WS-Addr Service <ReplyTo> Consumer Callback Portal 1 2 3 4 5 WS-Addr <ReplyTo> BAM CEP Callback ERP 1 2 3 4 5 CRM Rules Web Service Provider 29
  29. 29. Service Component Architecture (SCA) Combine BPEL, BAM, Rules, Custom Logic… Web WS-Addr Service <ReplyTo> Consumer Callback Portal 1 2 3 4 5 WS-Addr <ReplyTo> BAM CEP Callback ERP SCA Composite 1 2 3 4 5 CRM Rules Web Service Provider SCA Composite 30
  30. 30. Agenda/Outline • SOA Today: Drivers for Change • SOA Tomorrow: SOA Grid • State Management in the SOA Grid • SOA Grid – Advanced Capabilities • QoS and Distributed SOA Processing • Relocatable Stateful Orchestrations (BPEL) • Not Your MOM’s Bus • New Model for Scaling SOA • Summary 31
  31. 31. DataCenter Resource Utilization • What Does it Mean to Scale a Service? • A New Model for Service Scalability • More Effective Service Virtualization 32
  32. 32. Scaling SOA - Approaching Limits and Bottlenecks • Memory, Disk, CPU • Typical Virtualization Solutions don’t optimize this balance • CPU is maxed, then memory is underutilized • Memory is maxed, then CPU underutilized • Disk bottlenecks can back up processing • Necessitates Queueing • Scale-down can kill sessions (Coherence Web solves this) • SOA Grid • Disk bottlenecks virtually eliminated • Memory utilization delegated to storage enabled grid nodes • Only CPU-bound resources need to be load-balanced. 33
  33. 33. Server Utilization Typical Virtualization Scenario Server A Server B Server C Memory Usage Memory Usage Memory Usage Memory 50 % 10 % 20 % CPU CPU Usage CPU Usage CPU Usage … 10 % 50 % 10 % Disk Usage Disk Usage Disk Usage Disk 20 % 20 % 50 % 34
  34. 34. Server Utilization SOA Grid Server Z Memory Usage Memory 50 % … CPU Usage CPU 50 % Disk Usage Disk 1% 35
  35. 35. Scaling SOA A new model for efficient resource utilization Data Memory Usage Data Memory Usage Data Memory Usage 50 % 50 % 50 % … Service A Logic Logic Logic CPU Usage CPU Usage CPU Usage XSLT Transformation … ESB 50 % 50 % 50 % Logic & Data Logic & Data Logic & Data Memory Usage Memory Usage Memory Usage Service B 50 % CPU Usage 50 % CPU Usage 50 % CPU Usage … Application Logic 50 % 50 % 50 % BPEL Process 36
  36. 36. SOA Grid • State-aware continuous availability for service infrastructure, application data, and processing logic • Predictable scalability • Scales out linearly, whether 2 or 2,000 servers • Heterogeneous Environment • High-end / low-cost commodity hardware • Dramatic overall increase in performance and throughput • Linearly scalable shared memory and logic • Reduced dependency on disk persistence • Reduced operations cost • Reduced deployment complexity • More efficient server resource utilization • Less flavors of servers to deploy 37
  37. 37. For More Information • SOA Magazine – SOA – Ready for Primetime: The Next Generation, Grid Enabled SOA • http://www.soamag.com/I10/0907-1.asp • Data Grid – • http://www.oracle.com/products/middleware/cohere nce/index.html • General SOA Information – http://www.oracle.com/technologies/soa/ index.html 38
  38. 38. Q&A 39
  39. 39. 40

×