The Anti-Pattern Wall of 2011 has Crumbled: Stateful Applications in Containers
summary
The 2011 rule that apps must be stateless is no longer absolute: container runtimes plus orchestrators now host stateful services using Kubernetes StatefulSet, operators, Helm charts, DC/OS/Mesos frameworks and external persistent volume mounts. Docker Hub showed 7 of the top 12 images were stateful when sampled. Practical tooling includes Rex-Ray and a draft Container Storage Interface for portable drivers; snapshots and standardized plugins remain work in progress. Production usage exists (Kafka, Cassandra, Postgres variants) with varying tolerance for failure.
AI-written summary · about
chapters (17)
- 0:00Start
- 1:54Container History Lesson
- 4:1512 factors published
- 7:16PaaS vs a Container Orchestrator
- 9:01What does the 12 factors say?
- 11:11What exactly is this stateful backing service? In the era when the twelve factors was written it was typically a database.
- 12:25What if it isn't the other guys problem? Suppose you are that guy maintaining the backing store There are valid reasons to do this yourself
- 14:00Monolithic = bad
- 21:47local storage for state?
- 23:11How External Persistent Volume Mounts work
- 26:59Kubernetes Helm Chart
- 30:58Demo
- 31:40The 12 factors revisited
- 32:47What needs improvement in the stateful story Backup is ripe for standardization
- 37:12Stateful app support is under construction
- 40:01Technology adoption is nonlinear
- 41:32NYC 1913 - spot the horse
related talks
37mStateful vs Stateless Container Patterns in Kubernetes, a Deeper Analysis
38mThe Container Operator’s Manual
29mStateful Workloads in Kubernetes: A Deep Dive
29mUnderstanding the Container Revolution
30mContainers and Next Gen Infrastructure Ecosystem
01h 04mLaunching patterns for containers - it's more than just scheduling
43mContainers Change Everything
37mBuilding Stateful Workloads in Kubernetes