+ All Categories
Home > Technology > StackEngine Problem Space Demo

StackEngine Problem Space Demo

Date post: 02-Jul-2015
Category:
Upload: boyd-hemphill
View: 318 times
Download: 7 times
Share this document with a friend
Description:
Docker Adoption comes with a set of problems for productions operations. Tools simply do not exist for the problems a Not Pets, Not Cattle but Ants application topology comes with. StackEngine demos its Beta release in the context of each of these problems and how it proposes to address them.
47
Managing Docker Applications in Production Boyd E. Hemphill @behemphi @stackengine
Transcript
Page 1: StackEngine Problem Space Demo

Managing Docker Applications in Production

Boyd E. Hemphill @behemphi @stackengine

Page 2: StackEngine Problem Space Demo

History - BoydOracle Developer, MySQL DBA, PHP Developer, System Architect, DevOps Director (yea, really)

Started Austin DevOps In 2012

Co-Organizer of DevOps Days Austin

Became involved in Docker Austin in August.

Working with Docker for the first Docker Days Austin

Keep an eye on @stackengine to see if we can make it happen!

Page 3: StackEngine Problem Space Demo

History -StackEngine

CopperEgg

Started in May 2014 to address problems arising due to Docker adoptions

Funding announcement October

Currently in private beta

[email protected]

Page 4: StackEngine Problem Space Demo

Common Ground

Page 5: StackEngine Problem Space Demo

What is The Goal of Your

Company?

Page 6: StackEngine Problem Space Demo

What is The Goal of Your

Company?Make Money!

Page 7: StackEngine Problem Space Demo

So … What is DevOps?

Page 8: StackEngine Problem Space Demo

Is DevOps a Process?

Page 9: StackEngine Problem Space Demo

Is it an intersection of overlapping concerns?

Page 10: StackEngine Problem Space Demo

Is DevOps a Culture?

Page 11: StackEngine Problem Space Demo

DevOps:

DevOps is the way in which a technology organization embeds itself in a business to the benefit of that business.

Page 12: StackEngine Problem Space Demo

The Ideal Docker Application

Page 13: StackEngine Problem Space Demo

Netflix (unicorn)Service Oriented Architecture

Page 14: StackEngine Problem Space Demo

BenefitsIncrease developer velocity using containers to model application topology

Decrease build and test times using containers as build/test agents rather than VMs

Density - Reduce Costs

Separation of Concerns - “Code to an interface” moves to the infrastructure level

Taken together a company that can get to the end state has a significant and lasting competitive advantage

Page 15: StackEngine Problem Space Demo

Today’s Docker Reality

Page 16: StackEngine Problem Space Demo

ConcernsThis is still very early days for Docker

Tools to support Dockerized applications immature or do not exist

Service oriented architecture is hard

Increasing developer is only a local optima, thus a false economy

Its the first major DevOps initiative not driven by Ops.

Page 17: StackEngine Problem Space Demo

SDLC

Developers are moving forward

QA sees production modeling and large test grids

Build engineers see easier build grids

Operations panics - no tools or instrumentation

Page 18: StackEngine Problem Space Demo

The Final Mile

Page 19: StackEngine Problem Space Demo

The Final Mile

Developer uses containers

System is built and images are produced

Images stored in a private repo

Now we need to deploy to a QA environment

Umm ….

Page 20: StackEngine Problem Space Demo

Docker DeploymentCompleting the final mile

Page 21: StackEngine Problem Space Demo
Page 22: StackEngine Problem Space Demo

DeploymentContinuous Deployment was published in 2010

In general we still talk about build and integration, not deployment

This is true for all applications, not just those that are Dockerized

Docker promises A/B deployment

StackEngine is working on making A/B deployment a policy based activity

Page 23: StackEngine Problem Space Demo

Container Sprawl

Page 24: StackEngine Problem Space Demo

Container Sprawl

First there was VMware and suddenly a large corps departments could easily spin up environments

Then there was the cloud and developers could complete circumvent Ops

Now there is Docker and Developers are creating numerous abstractions that are not managed like traditional VMs

Containers on VMs on phyical hardware located in availability zones in data centers in geographic regions.

Page 25: StackEngine Problem Space Demo

My Head HurtsContainer Sprawl

Page 26: StackEngine Problem Space Demo

Inventory Management

Page 27: StackEngine Problem Space Demo

Where’s My StuffHosts

Page 28: StackEngine Problem Space Demo

Where’s My StuffImages

Page 29: StackEngine Problem Space Demo

Where’s My StuffContainers

Page 30: StackEngine Problem Space Demo

Telemetry

Page 31: StackEngine Problem Space Demo

Telemetry

What does monitoring a docker app look like?

What does log aggregation look like?

Page 32: StackEngine Problem Space Demo

Telemetry

Container size matters - blog

Log aggregation solution - blog

Page 33: StackEngine Problem Space Demo

What’s my stuff (not) doing

Containers

Page 34: StackEngine Problem Space Demo

What’s my stuff (not) doing

Containers

Page 35: StackEngine Problem Space Demo

What’s my stuff (not) doing

Containers

Page 36: StackEngine Problem Space Demo

Components

Page 37: StackEngine Problem Space Demo

Components

Components are infrastructure abstractions (cloud provider, data center, docker registry, docker image)

Components can be arranged in logic ways to represent physical infrastructure.

Page 38: StackEngine Problem Space Demo

Composing and AppComponents

Page 39: StackEngine Problem Space Demo

Composing and AppComponents

Page 40: StackEngine Problem Space Demo

Composing and AppComponents

Page 41: StackEngine Problem Space Demo

Service Discovery

Page 42: StackEngine Problem Space Demo

Service Discovery - Consistency Models

Zookeeper

Etcd / Consul

Chef Server / Heira

Page 43: StackEngine Problem Space Demo

Service DiscoveryDue out in December

Allows multiple deployments of the same service on the same host.

Containers register themselves upon startup and are found by tags rather than URI’s, fixed ports, etc.

A step towards cattle containers instead of pet containers

Page 44: StackEngine Problem Space Demo

Roadmap

Page 45: StackEngine Problem Space Demo

Roadmap

Service Discovery

User Management

Addition of new Components

Better telemetry

Scheduling

A/B Deployment support

Page 46: StackEngine Problem Space Demo

Live Product DemoAl Gore and his gremlins willing

Page 47: StackEngine Problem Space Demo

Boyd Hemphill @behemphi

@stackengine

Thank You for Your Time and Comments.


Recommended