DevOps in Real Life, A How-To
summary
Meltwater builds a central platform team that offers a Drone CI pipeline, self‑hosted Kubernetes via kops, auto DNS, and templates so teams push code, get auto‑deployed to staging/production, and inherit logging and metrics. Metrics run on Prometheus + Thanos with Grafana; logs flow from Filebeat→Logstash→Elasticsearch/Kibana; the logging cluster holds ~240 TB with ~300k log events/sec. Adoption scales from 2 pilot teams to ~30 teams and hundreds of nodes; platform handles upgrades, node failures and frees teams to focus on business logic.
AI-written summary · about
chapters (22)
- 0:00Start
- 3:45DEVELOPERS | OPERATIONS
- 7:17EMPOWERING TEAMS
- 8:15SCARY
- 8:39Deliver innovative features catered to the customers needs
- 9:15INTRODUCING THE PLATFORM TEAM
- 10:43SO YOU JUST CHANGED NAME OF THE OPS TEAM TO PLATFORM TEAM?
- 10:57WHERE ARE YOUR BOUNDARIES OF RESPONSIBILITY?
- 20:23CONVENTIONS
- 27:06BUY VS BUILD
- 27:58AUTONOMY
- 28:59PLATFORM DELIVERS A PRODUCT
- 31:11CONTAINER ORCHESTRATION AS A SERVICE
- 31:34kubernetes
- 32:25CLUSTER MANAGEMENT
- 35:28AUTOMAGICAL LOGS HARVESTING
- 37:35AUTOMAGICAL METRICS HARVESTING
- 42:18COLLABORATION
- 43:22CONTINUOUS IMPROVEMENT
- 44:13LESSONS LEARNED
- 45:52CULTURE
- 47:26QUESTIONS?
related talks
48mDevops@Scale
28mDevOps meets reality
52mDevops @Scale
28mDevOps in a Data Warehouse: Inside Out
45m200 Teams, 1 Goal: DevOps Maturity at Scale
58mDevOps @ Scale — Greek Tragedy in Three Acts
55mDevOps @Scale (Greek Tragedy in 3 Acts)
40mDevOps is for startups!Enterprises do platform engineering and here is why - Jürgen Süßner