Share
Embed
Building a Scalable Python gRPC Service using Kubernetes
Ian Lewis
June 17, 2015
Transcript
36 slides
-
Building a Scalable Python gRPC Service using Kubernetes @ PyCon
APAC Ian Lewis, Developer Advocate -
gRPC Concepts gRPC & Python Containers & Kubernetes Concepts Running
a Kubernetes Service Wrap Up 1 2 3 4 Agenda 5 -
Section Slide Template Option 2 Put your subtitle here. Feel
free to pick from the handful of pretty Google colors available to you. Make the subtitle something clever. People will think it’s neat. gRPC Concepts Wait…. RPC? What happened to REST? -
• is binary ( no telnet :( ) • is
fully multiplexed (fewer connections) • uses header compression to reduce overhead • allows servers to “push” responses HTTP/2 ... -
• are a way to serialize structured data • are
binary ( really no love for telnet :( ) • can be parsed quickly • are flexible (easy to add new fields, supports addons) • support code generation Protocol Buffers ... -
message Person { required string name = 1; required int32
id = 2; optional string email = 3; } Protocol Buffers -
// The greeting service definition. service Greeter { // Gets
my Person data rpc GetMe () returns (Person) {} } Protocol Buffers -
Section Slide Template Option 2 Put your subtitle here. Feel
free to pick from the handful of pretty Google colors available to you. Make the subtitle something clever. People will think it’s neat. gRPC & Python Python! -
Section Slide Template Option 2 Put your subtitle here. Feel
free to pick from the handful of pretty Google colors available to you. Make the subtitle something clever. People will think it’s neat. Containers & Kubernetes Concepts What that thar be?!?! -
kernel libs app app app No isolation No namespacing Common
libs Highly coupled apps and OS app Shared Machines -
Some isolation Expensive and inefficient Still highly coupled to the
guest OS Hard to manage app libs kernel libs app app kernel app libs libs kernel kernel Virtual Machines -
build and deploy vm images • Curated runtimes • Rich
services • Auto-everything • … just add code • Managed collections • Declarative + Dynamic Compute as a Continuum Platform Cluster • Basic atom • Run anything VM More agility More Flexibility build and deploy containers build and deploy apps -
Everything at Google runs in containers: • Gmail, Web Search,
Maps, ... • MapReduce, batch, ... • GFS, Colossus, ... • Even Google Compute Engine: VMs in containers! -
Everything at Google runs in containers: • Gmail, Web Search,
Maps, ... • MapReduce, batch, ... • GFS, Colossus, ... • Even Google Compute Engine: VMs in containers! We launch over 2 billion containers per week. -
Why containers? • Performance • Repeatability • Isolation • Quality
of service • Accounting • Visibility • Portability A fundamentally different way of managing applications -
Greek for “Helmsman”; also the root of the word “Governor”
• Container orchestrator • Runs Docker containers • Supports multiple cloud and bare-metal environments • Inspired and informed by Google’s experiences and internal systems • Open source, written in Go Manage applications, not machines -
• 0.18.0 released • 369 contributors • over 8000 github
stars • CoreOS • HP • IBM • Mesosphere • Microsoft • Pivotal • Red Hat • SaltStack • VMWare • Binary releases • Logging and monitoring • Richer scheduling ◦ Labels based constraints ◦ Smart bin-packing ◦ Run-to-completion • Updated service model ◦ IP per service • DNS integration Project Partners New Introducing Google Container Engine Community -
• Group of containers • Settings in a template ➔
Reuse across environments ➔ Repeatable, manageable Behavior Benefits Pod Web Server Content Management Server Pods more at pod.md -
Dashboard show: FE Labels • Metadata with semantic meaning •
Membership identifier ➔ Allow for intent of many users (e.g. dashboards) ➔ Build higher level systems … Behavior Benefits Pod Pod frontend Pod frontend Pod Pod FE BI, FE v2 Dashboard show: v2 more at labels.md -
Replication Controllers Behavior Benefits Replication Controller #Pods → 2 label
selector: v1 Pod Pod frontend Pod frontend Pod Pod v1 v1 Replication Controller #Pods → 1 label selector: v2 v2 more at rc.md • Keeps Pods running • Gives direct control of Pod #s ➔ Restarts Pods, desired state ➔ Fine-grained control for scaling -
Pod Services Behavior Benefits Pod FE FE Pod FE Service
label selector: Front End more at services.md ➔ Clients shielded from implementation details ➔ Independently control each, build for resiliency • Stable address • Decoupled from Controllers -
Section Slide Template Option 2 Put your subtitle here. Feel
free to pick from the handful of pretty Google colors available to you. Make the subtitle something clever. People will think it’s neat. Running a Kubernetes Service Steer me ship captain!