APM in the cloud: Are you ready? THANK YOU By: Mike Taylor Compuware Confidential. Do Not Duplicate The value of customer satisfaction Average impact of 1 second delay in response time for Web users Page Views Conversions • • • • • Customer Satisfaction Lost revenues* Brand damage More support calls Increase costs LOB dissatisfaction with IT *Online business doing $100K/day = $2.5M/year in lost revenue Macro Trends Demand A New Approach To APM Complexity Explosion Business Demands Increasing More features! Faster delivery! Faster performance! User Expectations Continue to Rise Do More With Less Workload Staff and budget 3 APM Solutions Application-Centric World Application Performance Management Challenges 1. Technology 2. People and Approach Business Development Test Production C/C++ Application Performance Management Challenge #1 1. Technology C/C++ Applications have become: • mobile and distributed • reliant on third parties • cloud-based • increasingly complex and fragile Modern Application Delivery Chain The Application Delivery Chain Data Center Cloud: Private and Public Web ▪ Mobile ▪ App logic ▪ Database ▪ Network ▪ Mainframe ▪ Virtualization ▪ SOA ▪ CDNs ▪ Third party services Users ISPs ▪ Mobile carriers ▪ Browsers ▪ Devices ▪ AJAX ▪ JavaScript ▪ Mobile apps Customers Application Employees Infrastructure Traditional Application Performance Management Traditional Approach …user is happy DB Network Server Data Center The Application Delivery Chain Cloud: Private and Public Web ▪ Mobile ▪ App logic ▪ Database ▪ Network ▪ Mainframe ▪ Virtualization ▪ SOA ▪ CDNs ▪ Third party services Users ISPs ▪ Mobile carriers ▪ Browsers ▪ Devices ▪ AJAX ▪ JavaScript ▪ Mobile apps Customers Application Employees Infrastructure The Application Performance Challenge: Traditional Solutions are Insufficient ! ! Partial coverage = The Application Delivery Chain No visibility into users’ blind spots and no view experience and related of complete app business impact Customers ? ? ? ? ? Application Infrastructure ! Employees Focus on infrastructure, not application The Answer: Adopt an Application Point of View That Starts with the User Application Point of View that Starts with the End User Data Center Cloud: Private and Public Web ▪ Mobile ▪ App logic ▪ Database ▪ Network ▪ Mainframe ▪ Virtualization ▪ SOA ▪ CDNs ▪ Third party services Users ISPs ▪ Mobile carriers ▪ Browsers ▪ Devices ▪ AJAX ▪ JavaScript ▪ Mobile apps Customers Application Application Employees Infrastructure Appliance based Solutions • Production Operations focused • Great breadth of technology • fault domain isolation 11 Agent Based Technologies • Application focused • depth • root cause analysis 12 Application Performance Management Challenge #2 2. People and Approach Issues: • Business demands are increasing • Applications are changing rapidly • Too many dev/test iterations • Too long to deliver and repair Business Development Test Production Simple View of Typical App Performance Lifecycle Business Development (local, remote, outsourced) Test/QA (local, remote, outsourced) • Load testing Production (local, remote, outsourced) • Cloud load testing • Monitoring Problems with Typical App Performance Lifecycle ✘ Business impact? ✘ Priority? ✘ Competitive info? Business Not enough business context! $$$$$$ ✘ What? ✘ Who? ✘ When? ✘ How? ✘ Code? ✘ Recreate? Development (local, remote, outsourced) Too much time reproducing problems! Test/QA (local, remote, outsourced) • Load testing Not engineered for performance! Too many iterations! Production (local, remote, outsourced) • Cloud load testing • Monitoring Too many business processes impacting issues! Lifecycle-Oriented APM: Single System Which users $$ amount Conversions Abandonment Etc. Business Business impact $ All transactions Click-to-code All details Development (local, remote, outsourced) No need to reproduce issues Test/QA (local, remote, outsourced) • Load testing Performance from the start Fewer iterations Production (local, remote, outsourced) • Cloud load testing • Monitoring 24x7, all transactions Fewer issues APM Should Address the New Application Challenges 1. Technology 2. People and Approach Business C/C++ C/C++ Test Unified coverage of application delivery chain Development Production Lifecycle by design APM Options in the cloud dynaTrace APM: Howinit the Works 1. Appliance based APM solutions cloud Private or Public Cloud Hosted by Vendor Browser / Rich-Client Synthetics Web Server Java .NET VB, C++, Other Database Mobile Probe Can you guess the Benefits and challenges Appliance based APM solution provided by vendor usually Server Corporate Network Client Great application Tx and network visibility Limited in deep dive in applications dynaTrace How it Works Private/Public cloud dataAPM: on Corporate Network Private or Public Cloud Hosted by Vendor Browser / Rich-Client Synthetics Web Server Java .NET VB, C++, Other Mobile Easy to deploy and maintain SaaS Vendor Corporate Network Client Sampling based and limited transaction tracing Server Compuware Confidential. Do Not Duplicate Database 2. SAAS Based APM Private or Public Cloud Hosted by Vendor Browser / Rich-Client Synthetics Web Server Java .NET Mobile Collector Information is extremely detailed Root cause analysis is very easy Dashboard configuration requires planning Corporate Network Client Server Server VB, C++, Other Database dynaTrace it Works Private/Public cloud dataAPM: in theHow cloud Private or Public Cloud Hosted by Vendor Browser / Rich-Client Synthetics Web Server Java .NET VB, C++, Other Mobile Server Collector Servers Provided and maintained by cloud provider Corporate Network Client Detailed client information kept in the cloud Compuware Confidential. Do Not Duplicate Database dynaTrace APM: How it Works Content Delivery Networks CDN Browser / Rich-Client Corporate Network Web Server Java .NET VB, C++, Other Client Synthetics Collector Mobile CDN • • Database Dynamic content • Java script injection • Third party content CDN response time metrics Compuware Confidential. Do Not Duplicate Server BigData Awesomenss End-to-End Put Performance APM and Business into Context End User Web Server User Experience Java .NET Other RDBMS & NoSQL Big Data APM Monitor, Troubleshoot, Optimize Server 100% all Users Business Analysis Clean Business Data BigData High Volume/Low Latency DBs Web Key Challenges All typical Database Challenges Even Distribution Correct Schema Understanding Application Impact No established DBA/Monitoring Tools 25 Java Key Benefits Fast Read/Write Redundancy and Scalability Hadoop: Large Scale Parallel Processing Master Node Hadoop Cluster Job Tracker Name Node Key Challenges 1) Optimal Distribution 2) Unwieldy Configuration 3) Can easily waste your resources 4) Failure or Error Analysis is hard 5) Performance Optimization is hard 1 Data Node per Host 1 Task Tracker per Host Many Task JVMs per Host 1) HDFS: Distributed File System 2) MapReduce Key Benefits 1) Massive Horizontal Batch Job 2) Split big Problems into smaller ones 3) AdHoc Analytics of 100s of Terrabytes