Deploying TeamCity Cloud as a Service on AWS: Lessons Learned
summary
Each customer runs a per-client isolated TeamCity instance (ECS container), artifacts on S3, metadata split per-schema in RDS (MySQL), and shared state on EFS; networking lessons: NAT gateway costs and unpredictable egress lead to using VPC endpoints and NAT instances and careful CIDR sizing. Hit operational limits: ALB rule limits and throttling from ECS health checks, EFS throughput exhaustion on simultaneous starts, logging locks when AWS log driver blocks, and IAM role-count scaling pain — prompting an ECS→EKS migration for more control.
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:24TC Cloud intro: history, TC in general
- 5:21TC Cloud architecture overview
- 6:43Managing TC Cloud
- 8:56Lessons learned: VPC, subnets
- 11:56Lessons learned: ALB
- 13:38Storage: EFS, S3
- 15:45Elastic Container Service
- 18:15Identity Access Management
- 20:30RDS
- 22:31General recommendations
- 26:24Q&A
related talks
25mGetting started with TeamCity Cloud
28mSaaS migration and modernisation
34mLessons Learned from Five Years of Multi-Cloud
45mSecurity Lessons Learned Moving a Bank 100% in the Cloud
52mBuilding and running Spring Cloud-based microservices on AWS ECS
35mMigrating Your Old Server Products to Be Stateless Cloud Services
32mCreate a cloud-ready environment
49mImplement, scale, operate & innovate: TennCare’s cloud strategy