Day2 のメモ。
Keynote Bringing the Cloud to You
- G Suite
- BigQuery ML
- Cloud Service Platform
- Cloud Armor
- Edge TPU
Microservices - Born and Raised (and Growing) on Google Cloud
Pizza Hut の実例。
- Kubernetes
- gRPC サービスと OpenAPI でアグリゲーション
- HA (Global Load Balancer)
Next
- istio
- apigee
API Management Best Practices
- Measuring Value to the End User
- Application, Platform, APIs Usage
- Traffic Growth
- Measuring Value to the Application Developer
- #Consuming Apps
- #Developers
- Community Growth
Where Should I Run My Code? Serverless, Containers, VMs and More
- It depends
- コンテナでレイヤの移動は容易になる
- 上に行くほど考えることは減る
GCP | fit | constraints | hybrid |
---|---|---|---|
GCF | serverless, pub/sub or storage, don't want to think about runtime env, data transformations (ETL), and cloud-based http glue | runtimes (lang), function level granularity, and must interact via events | Function as a Service |
App Engine | http/s, stateless, and scaling | runtimes (lang and binary extension), inherit docker constraints | Envoy, Istio & Knative on k8s |
GKE | consistency, containers, team communication (dev, ops, security, etc.), ci/cd pipeline, and beyond http/s | must use containers, licensing, some architectual constraints | k8s |
GCE | existing systems, specific os/kernel, databases, and beyond http/s | scaling speed ~20s per vm, 1000s in ~60s, updates yourself | Virtual machines |
From Your Workstation to Production: The Serverless Development Lifecycle
- code
- typed
- testing
- deployment/rollback
- cloud build
- config/environments
- monitoring
- stackdriver
- pricing
One Platform for Your Functions, Applications, and Containers
see also 108
Knative
- Serving
- blue/green
- auto scaling
- auto builds
- Eventing