Encontrando la Aguja en el Rendimiento de Aplicaciones
0 likes875 views
The document discusses the challenges of building, testing, and deploying high-quality applications quickly in a complex, hybrid software environment. It highlights the importance of managing various application components and end-user experience, along with the need for effective diagnostics and performance monitoring to isolate issues and ensure security. Key features include tools for problem isolation, real-time performance monitoring, and customizable dashboards to aid in application development and testing.
Encontrando la Aguja en el Rendimiento de Aplicaciones
- 1. Encontrando la Aguja en el Rendimientos de Aplicaciones. Octubre 2015
- 2. It’s not always this easy…
- 3. Innovating faster… Every business is a software business today! This means building, testing and releasing quality software faster and more frequently.
- 4. Applications are transforming Hybrid composite applications HTML5 / streaming protocols / social media Mobile explosion Rich internet apps Hybrid delivery Composite apps SOA / private cloud / middleware / mainframe CDNs / public cloud Web and native / myriad of devices Web | Native | Hybrid
- 5. The challenge: Building high quality apps at the speed of business How do I deploy apps faster? How do I measure how end users are experiencing my apps ? How do I ensure my app is secure and reliable To minimize business risk? With today’s competing priorities, how do I prioritize my investments and planning? How do I decide what to test, how to do it continuously, and manage it all? How do I isolate problems faster?
- 6. What is Composite Application Management? A combination of legacy, packaged and new application logic developed on a J2EE/.NET and middleware/legacy platforms (Gartner) More than 80% of enterprise business processes traverse through composite applications The future of application architectures as SOA is accelerating the adoption and complexity (Gartner) Source: IBM End users HTTP, JMS, RMI, MQ HTTP, JMS, RMI, MQ JDBC, PL SQL HTTP, JMS, RMI, MQ MQ MQ MQ What seems to be the problem?
- 7. Managing Complex Environments End User Experience including: • DNS Time • Network Time • Client and Download Time • Server Time • Error Screenshots • Real and Synthetic Users Client • Process Orchestration (UML, BPEL, etc) • Transaction Routing • Messaging (JMS, MQ, RV, etc) Middleware • Web and App Server • Portals • Presentation Layer • Business Transactions • Business Logic • Tiers • Components • Methods Web and application servers • Mainframe CICS, IMS, Batch, DB2 • Enterprise Applications: Oracle, SAP, PeopleSoft, Siebel • Oracle 8,9 & 10g DB Back-end systems and applications End-to-end transaction management
- 8. PlanningPlanning App Development App Development App TestingApp Testing App release App release Release decision Release decision Rapidly increasing WIP Lack of effective customer insight and high latency drives “kitchen sink” requirements Waiting time for build and test environments drives “desk-side” builds Poor confidence in test data fosters “release aversion” driving more WIP Isolated build and integration processes Deployed App Deployed App Manual Testing increases latency or drives limited test coverage Manual and error prone app deployments Error prone manual hand-offs and processes High # defects Locally optimized teams, measures and tools drives escalating WIP, lack of end-to-end visibility and trust One way flow Poor user experience “patch in production” leads to snowflake systems InfoSec & compliance engaged late driving vulnerabilities & re-work High velocity innovation, Business demands Business demands Continuous assessment continuous delivery Build Test Release Deploy Operate Monitor Continuous assessment Continuou s Operations Continuous Delivery & Deploymen t Continuou s Integratio n & Testing
- 9. Transaction Tracing Database App Server Time-VMs • Tier • Component • Method • Outbound Resource Database Time • JDBC • Connect • Execute • SQL Query Web Server Time • Servlet/JSP • ASP • Method • Struts • Portals • WebService WebServer Transaction “A” Client Request Transaction Header APPLICATION APPLICATION SERVER VM2 EVENTS APPLICATION APPLICATION SERVER VM EVENTS Client Request Transaction Header Client Request Transaction Header How does it work? Cross VM • Webservice • HTTP • RMI/T3/T3S • JMS/MQ
- 10. Diagnostics can help you answer… • Which components are associated with my applicationtransactions? • Which services and infrastructure does my application depend on? • Why are my transactions slow? • How can I find the root cause for application slowdown? • How can I find what is causing my application server to crash or to fail processing transactions? • What changed between today and last week? • What is hiding behind “server time” which shows up in my EUM reports? • How can I give the application team actionable information to investigate complex problems?
- 11. Diagnostics Value Proposition Quickly find the right hay stack and make it easy to find the needle.
- 12. Problem isolation Is there a bottleneck?
- 13. Is there a Memory leak? Problem isolation
- 14. Actionable Information Where is the bottleneck? What component? What object? What method? What does a call look like before, during and after the problem manifested?
- 15. Flow Tracing & resource based utilization How does an issue impact the business and other systems What is the business impact of this issue? What other applications using shared components might be effected? Which component’s expert team(s) need to be involved?
- 16. Investigate Component Load View components load and investigate further View components load and investigate further
- 17. Easy to Create Custom Dashboards Select entities and metricsSelect entities and metrics Drag & Drop existing views or easily create your own UI components Drag & Drop existing views or easily create your own UI components Easily create custom dashboards and share via MyBSM Easily create custom dashboards and share via MyBSM
- 18. Diagnostics for Performance Testers Monitor application performance during load testing and performance testing • Real time view of application performance with same Diagnostics interface used in production • Integration with HP LoadRunner and HP Performance Center provides post test performance analysis
- 19. LoadRunner / Performance Center real time integration App monitoring during load and performance test App monitoring during load and performance test
- 20. Feature: Diagnostics data in LoadRunner reports Diagnostics data from load tests is stored offline and can viewed in LoadRunner offline analysis tool Diagnostics data from load tests is stored offline and can viewed in LoadRunner offline analysis tool
- 21. Diagnostics for Developers Free Diagnostics Profiler • In-depth profiling for a single JVM • Allows remote access to application server through main Diagnostics UI Automatic leak pinpointing • Unique approach finds leaks with very low overhead Thread State Analyzer • Quick visual isolation of thread issues • Displays state distribution percentage by thread over time JUnit run comparisons • Show performance results in unit tests when resolving difficult performance problems
- 22. Intelligent Call Profile Instance Selection Select call tree instances for future inspection Select call tree instances for future inspection
- 23. Tracing Web Services Across Tiers Cross VM activity is correlated and shown in a single call profile Cross VM activity is correlated and shown in a single call profile
- 24. Tracing Messaging Activity Across Processes Correlate JMS activityCorrelate JMS activity
- 25. Superior Visibility Into Exceptions Get visibility into exceptions thrown, even if those are reported as part of a cross-VM call Get visibility into exceptions thrown, even if those are reported as part of a cross-VM call
- 26. Catch unexpected declines in typical user activity Catch unexpected declines in typical user activity Constantly compare expected and actual user behavior Constantly compare expected and actual user behavior Alert on a performance regression after a new release Alert on a performance regression after a new release Diagnostics – Predictive Analytics
- 28. ERP/CRM Services Database End Users Legacy Services Web Service Clients Firewall BSM BAC Diagnostics Applet LR/PC Application Server 1 Portal Servers Application Server 2 Commanding Server Collector Process Distributed Server Architecture
- 29. © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. Thank you¡Diagnostics Architecture!