Deploying and maintaining software for today's infrastructure
summary
Panelists report concrete practices and results: spot instances cut costs and teams spend two to three days to prototype infrastructure choices rather than months; clusters run 70–80 microservices in some orgs. Tools called out are Rancher for orchestration, wrappers that mimic cloud services (S3/DynamoDB) for local testing, and MapReduce-style stacks for big data. Hard lessons: proactively kill unresponsive Docker containers to avoid cascade failure, set resource limits per container, and replicate production topology locally to catch behavioral differences.
AI-written summary · about
chapters (11)
- 0:00Modern software deployment
- 3:12Infrastructure costs myth
- 6:11Innovation vs operational costs
- 9:30Infrastructure first principles
- 12:19Cloud trade-offs
- 17:43Operations culture shifts
- 23:41Deployment tools and platforms
- 29:39Dev/test environments cost
- 34:51Monitoring containerized systems
- 39:55Containers lifecycle
- 43:33Containerization adoption challenges
related talks
46mPanel Discussion: The number of orche
40mAgency Panel Discussion
30mBuilding Applications in the Cloud: Best Practices
36mPractical Cloud Native: What Works, What Doesn't
48mMoving from spaghetti infrastructure to composable cloud environments
42mArchitecting for Reliability
10mReducing infrastructure cost during development and in production
44mContain yourselves