Building Successful Apps Using Application Telemetry and Data Driven Decisions - Rakefet Jackoby Galazan, Microsoft - DevOpsDays Tel Aviv 2016
Download as PPTX, PDF0 likes307 views
Building successful apps using application telemetry and data driven decisions involves collecting telemetry from various tiers including the server backend, middleware, web services, and browser. This telemetry is processed and stored in an application performance management (APM) product. The APM product identifies, understands, and resolves issues using out-of-the-box and custom telemetry. It also monitors user behavior, application behavior through service dependencies and performance, developer traces, and infrastructure performance.
Building Successful Apps Using Application Telemetry and Data Driven Decisions - Rakefet Jackoby Galazan, Microsoft - DevOpsDays Tel Aviv 2016
- 1. Building successful apps using application telemetry and data driven decisions It’s time to t Rakefet Jackoby Galazan Senior Product Manager
- 2. It’s tough out there…
- 5. Telemetry is collected at each tier: server backend, middleware, web service & browser Telemetry arrives to the APM product where it is processed & stored Identify, understand and resolve issues with powerful out-of-the- box and custom telemetry
- 6. 1 Outside-in monitoring URL pings and web tests from different global points of presence Observed user behavior How is the application being used?2 Observed application behavior Service dependencies, queries, response time, exceptions, logs, etc. 3 Developer traces and events Whatever the developer would like to send to the APM product 4 Infrastructure performance System performance counters5
- 7. Plan 1 Monitor + Learn ReleaseDevelop + Test 2 Development Production 4 3
- 8. Manage work Develop + Test 1 Plan Project starts PlanTrack progress Prioritization and validation of investments not based on real data
- 9. Write Code Unit Testing 2 Build Version Control Build Verification Release Once the iteration starts, developers turn great ideas into features… Develop + Test Lack of collaboration between development and operations
- 10. Load Testing Integration testing environment Automated functional testing environment 3 Pre-production environment Staging environment Monitor + Learn Release Inability to quickly detect, diagnose and triage application issues
- 11. 4 Monitor Feedback Plan the next iteration
- 13. Machine learning algorithms for automatic detection Find disruptive trends and anomalous segments Correlated data for triage and fast diagnostics Configuration/customization may or may not be required
Editor's Notes
- #3: Building successful apps is hard! Whether you build a service or an app, there is fierce competition, and success requires data-driven decision making. It requires detecting and resolving issues before your customers do, and knowing how your product is being used to decide where to invest next
- #4: Modern Apps are constantly evolving , they are becoming more complex – they are built using multiple components, with growing dependency on first/third party integrations Apps are deployed in multiple DCs much faster than before Users are spread all across the globe using different clients and browsers. With Growing adoption of DevOps continuous delivery methodology there is an increasing amount of features that are deployed in very short timeframes. , With pressure to reduce the amount and lengths of slowdowns and outages there is a need to improve visibility and control of performance
- #5: And in order to stay on top of things, you need to be able to answer these questions, and many others, at all times, and in real-time.
- #6: So here is where APM products comes in. These products are all about helping make better apps . Ingestion, processing and visualization : By collecting data from all app tiers : backend , servers , web services & browsers, and providing powerful out of the box telemetry views of correlative data , APM products enable the users to detect triage and diagnose issues quickly , making the app more available and performant, with an acceptable user experience
- #8: So let’s get more concrete and talk about where in the app lifecycle an APM tool can help
- #13: As modern app developers, we all know how crucial it is to detect, triage and diagnose problems before they start affecting our customers. With APM you get all the tools to make your diagnostics experience smarter and find and fix problems before your customers know it! Detect: One of the most crucial things is to be able to detect issues as soon as they happen, and be alerted instantaneously. Triage: Once you detect an issue, the next thing is to figure out its impact and whether it is priority enough to solve right now. With APM you can find out the real user impact of any exception and take decisions accordingly. With an Application Map you can automatically detect your application topology across dependencies and client & server side components. You can find the impact assessment and click through to the resources to find the right information. Diagnose: Once you decide to fix an issue, you need all the context to solve it, and with the telemetry collection, you will have all the data you need.
- #14: But in the complexity of modern app architecture, even an army of analysts sitting in front of a dashboard cannot detect all the different things that can go wrong. That is where the more enhanced capability of proactive diagnostics come into play, which is one of the advanced capabilities that these products offer . Using Machine Learning based technology, you can be alerted on real time service disruptions and anomalous patterns in your app performance and behavior, with thresholds constantly evolving based on your app architecture and performance patterns.