Lost in Translation The Challenge of Managing Microservices Mirko - PowerPoint PPT Presentation

Lost in Translation The Challenge of Managing Microservices Mirko Novakovic CEO & Co-founder Instana

Innovate Fast Lines of Code Daily Builds 265,060,408 4,362 or Die Slowly: 150.00 “The digital transformation is fundamentally changing the way our consumers behave and the way we work. Technology has enabled us to build more direct relationships with our consumers… but also to become faster, better and more efficient in every part of the organization.” 33.85 2015 2016 2017 2018 2019

3 Tectonic Shifts in Application Operations Are Impacting Businesses Deployment DevOps Technical Frequency Change Organisation Complexity - Cloud/Hybrid - Continuous Integration - Value Stream Focus Deployment Automation - Microservices / Containers / - Sofuware Delivery Performance Service Mesh - Automated Testing - Agile - Serverless - Canary/Blue-Green Deployments - Trunk Based Development

Observability: Metrics, Logs, Traces?

Let’s have a look at a Microservices 731 Hosts environment 14 Containers / Host 731 Hosts 14 Containers / Host

Adding Kubernetes Clusters Namespaces Deployments Services Pods...

Services and Dependencies 940 Microservices 14 Containers / Host

Traces & Calls 731 Hosts 14 Containers / Host

Let’s do the math... 731 Hosts x 20 Metrics = 14,620 Metrics 731 x 14 Containers = 10,243 x 20 Metrics = 205,060 Metrics approx. 250,000 Metrics (not values) 50,000,000 Traces per Day 500,000,000 Calls per Day 100,000,000+ Logs per Day

To make it worse: Change! 731 Hosts 14 Containers / Host

Dashboards? Lost in Data...

Alerts? 731 Hosts 14 Containers / Host Clusters Namespaces Deployments Services Pods... Nam/ Host

New Challenges Releases occur so fast it’s It’s nearly impossible to find difficult to tell which ones your services in this chaos make things better or worse Impossible to make New services are sense of all of this data deployed without - Dashboards can’t help monitoring You can’t tell which service Finding the root cause of failed first and is breaking issues just got exponentially other services more difficult

How to get from data...

...to information?

4 Layers of Modern Monitoring DevOps Applications UEM Infrastructure CI/CD & Traces inc. Mobile & Technologies Information Intelligence Area of focus and competition Semantics Data Metrics, Logs, Traces (commoditized)

Dynamic Graph Connects Entities: - Traces, Calls - Services - Orchestration - Infrastructure - Cloud

Context Guide View upstream and downstream service Quickly see dependencies full stack dependencies Getting context in a complex and dynamic application landscape!

> - Application Perspective - Generated Dashboards - Full historical data - Filtered Services - Filtered Issues and Incidents - Corresponding Infrastructure - Drill into Analytics

Edmunds drastically reduced deployment time while increasing both quality and uptime Semantics: Releases, Deployments, Versions Get immediate, Commit Build Test Deploy automated feedback on DEVELOP every new release, deployment, and version with a full understanding Immediate visibility of the impact and changes and feedback

Intelligence Golden Signals: - High Level Abstraction - Learning/AI - Use Semantics for Context - Use Graph for Root Cause

Information: Root Cause Analysis Incident Mode - Full UI Rework - Focus on Incident in every Function of Instana - Playback - Dependency Map - More Event based ML

Information: Root Cause Analysis Outlier Detection - Slow Traces, Calls & Beacons - Erroneous Traces, Calls & Beacons - Depth of Traces, Calls & Beacons

Microservices Management Requirements Automated Discovery Immediate Feedback 731 Hosts & Monitoring on Delivery / Canary 14 Containers / Host Automatic & Assisted Root Cause Analysis Dependency Tracking & Visualization Developer Application Correlated > - Drilldown Analytic Analysis

Thank You