Why Your Team Can't Ship Fast Anymore
The Hidden Tax on Engineering Velocity
Austin Fahsl, Nx
Hi ๐Ÿ‘‹ I'm Austin
Sr Developer Productivity Engineer @ Nx with 10+ years of experience
Full-Stack Development
.NET + Angular, Enterprise & Startup
Open-Source Tooling
Lerna & Nx CLI
Developer Experience and Operations
Enhancing developer velocity with Nx Enterprise partners
The Same Questions, Over and Over
"Why do my automated checks take so long"?
"Why does my pipeline fail"?
"Why do I have to constantly fix my CI pipeline"?
"Why Do My Automated Checks Take So Long"?
What's Actually Slowing You Down
Sequential Execution
One task at a time
Redundant Work
Checks on unchanged code
"Longest Pole in the Tent"
Tasks you're always waiting on
Solution Phase 1: Foundational Setup
Intelligent Build System
Scalable Parallelization
Solution Phase 2: Continuous Improvement
Pipeline Metrics
Task Splitting
Company 1 - Accelerating Security Platform CI
Reduced CI Times from 2 Hours to Under 30 Minutes!
Filter (Nx Affected)
Replay (Nx Replay)
Distribute (Nx Agents)
"Why Does My Pipeline Fail?"
The Trust Problem
When your CI pipeline becomes unreliable, engineers give up on trying to fix it.
What's Actually Causing the Failure
1
2
3
4
1
Transient network errors or service outages
2
Dependencies on external environments
3
Legitimate build, test, or lint failures
4
Flaky end-to-end or integration tests
Solutions for Failing Pipelines
Smart Retry for Transient Issues
Mitigate impact of network issues, dependencies on other environments, and flaky tasks
Fast Iteration for Necessary Code Changes
Empower engineers to identify issues and implement code fixes quickly with the help of AI agents
Company 2: Stabilizing Accounting Platform's CI
Monthly Pipeline Failures Reduced by 37-50% with Automated Flaky Task Retries
"Why Do I Have To Constantly Fix My CI Pipeline"?
The Infrastructure Tax
Evolving Configuration
Fluctuating Resource Demands
Inefficient Parallelization
Solution: Smart Build System
Automatic Task Distribution
Dynamic Infrastructure Scaling
Company 3: Effortless CI Scaling for Vattenfall
Increasing the Maximum Concurrent PRs from 4 to 100!
No Bottlenecks in CI Resources
Simple CI Configuration
Adaptive Scaling on Demand
Chart Your Course - Ship Software Fast
Filter, Replay, Distribute
Use intelligent build systems and scalable parallelization algorithms.
Bolster Trust & Reliability
Leverage smart retries to stabilize pipelines and build confidence.
Simplify & Scale
Automate infrastructure to minimize maintenance and easily scale resources.
Thank You!
Nx & Nx Cloud @ nx.dev