DevOps 2.0 in hybrid clouds with Docker

Smart customization with no scripting. No cloud vendor lock-in. Full traceability

Overview

Frictionless #DevOps_2.0
Software for DevOps 2.0
Container services
Why Upgrade from DevOps 1.0
ARA / ARO Vendors
Technical Descriptions
No-script deploy?
References
More on DevOps

This article explains the quickest and least cost way to create servers for test or production use at enterprise scale.

Below is a summary of the funcamental strategies to achieve speed for less money. It’s not doing the same thing faster. It’s a new way of working that takes less time and less risk to achieve more.

Frictionless #DevOps_2.0

The narrative to the diagram above is here:

Servers are built on-premises or on major cloud environment (Amazon, Microsoft, Google, etc.), whichever cloud at the lowest cost. Such arbitrage can save money.
In addition to application programming source, infrastructure configurations are stored within a distributed versioned source repository (GitHub, or Bitbucket in the cloud or GitLab on-premises)
Server configurations are defined using declarative statements that specify the end result rather than specific steps This means when a server configuration is run several times, it doesn’t create a mess (this is called “idempotent”)
A single server configuration can be applied to one of several operating systems by use of Docker containers instead of hypervisors which repeat OS kernels in memory

Proprietary Hypervisors: Microsoft Hyper-V, VMware vSphere/ESi, Oracle OVM (Oracle Virtual Machine), Huawei FusionSphere

Open Source Hypervisors: KVM, Cisco Xen, Red Hat, OpenVZ
Automation engines (such as Shippable) covers the entire pipeline so integration of Jenkins with 3rd-party tools doesn’t risk projects
Differences between servers in test, demo, prod, and other stages are handled automatically without requiring explicit manual scripting Thus, monitoring and various other utility software for diagnosis are automatically added to each type of server
Instances can be created quicker if build results can saved in a build repository (such as Artifactory) for reuse rather than building each instance from source code

Common to both are notifications sent via variety of channels (email, SMS text, IRC, Slack, Skype, etc.).

TODO: graphic table like https://www.wikiwand.com/en/Industry_4.0

Software for DevOps 2.0

Joe Bada’s list is way more comprehensive, which this post highlights.

“DevOps 2.0” is achieved by tools from several developer groups.

GitHub / Bitbucket / GitLab
Docker.com
Kubernetes.com
Artifactory server image repository
Shippable.com
container registry services in cloud environments
Shippable Lighthouse service watches and sends notifications to Slack, IRC, etc.

Container services

Why Upgrade from DevOps 1.0

How long does it take to make a small change … and get it to the end customer? … through tests in various environments which gets more complex as more developers and components are added over time?

“DevOps 2.0” improves on where the “12 Factor App” promise of DevOps faces limits:

software too complex to learn
software too difficult to test in production
software too brittle
software operates only within a single vendor’s environment
inconsistencies of servers in different stages and environments

“Staging Servers Must Die” by Edith Harbaugh 2016 advocates for feature toggles – bits of code written into your application that let you turn features on or off at a moment’s notice for specific segments of your user base. It’s used by Facebook, Amazon. See Etsy’s Feature API on GitHub

There are also integration offerings such as Zapier, Microsoft Flow, etc.

“12 Factor App” is defined in 12factor.net and explained at:

Wikipedia)
In plain English
Building Microservices with the 12 Factor App Pattern on AWS slidedeck by Chris Hein
Applying the Twelve-Factor App Methodology to Serverless Applications, 12 FEB 2018 by Chris Munns
5 Years of 12 Factor Apps by Joe Kutner May 17, 2017 video explains it’s still as relevant as ever.
The Twelve-Factor App: What It Is and How to Monitor It Apr. 10th, 2018 by Adilson Somensari explains how New Relic helps with each factor.

ARA / ARO Vendors

Research firms Forrester calls it continuous delivery and release automation (CDRA) and Gartner calls it Application Release Automation (ARA)

Gartner predicts that “By 2023, 75% of global enterprises will have implemented at least one application release orchestration (ARO) solution, which is an increase from fewer than 20% today” in 2018.

ARA vendors are listed alphabetically below, along with Glassdoor.com reporting from employees who recommend the employer:

Wiki	Vendor	Glassdoor	Product/Solution	Docs
*	Atlassian	78%	-	-
*	BMC Software	68%	Release Lifecycle Management	*
*	CA Technologies	58%	CA Release Automation and Automic	-
*	Electric Cloud	58%	ElectricFlow Adaptive Release Orchestration	*
*	Flexagon	-	FlexDeploy	-
*	GitLab	81%	ARO	-
*	IBM	67%	IBM UrbanCode Deploy	-
*	Inedo	-	BuildMaster	-
*	Micro Focus	44%	Deployment Automation (formerly Serena)	-
*	Micro Focus	44%	Hybrid Cloud Management (Ultimate Edition)	-
*	Microsoft	81%	Azure Pipelines (Visual Studio Release Management)	*
*	OpenMake Software	-	DeployHub Pro	-
*	Octopus	56%	Deploy	-
*	Puppet	77%	Puppet Enterprise	-
*	Red Hat Open Shift	75%	OpenShift Container Platform and Ansible Tower	-
*	VMware	85%	vRealize Code Stream	-
*	XebiaLabs	97%	XL Release	-
*	XebiaLabs	97%	XL Deploy

ARA goes beyond just Jenkins, Ansible, and Continuous Integration.

ARA combines capabilities in automation, environment modeling, and release coordination for increased visibility for the whole team.

Technical Descriptions

Docker containerized environments use Kubernetes or Docker Swarm and Compose.

Kubernetes software watches and adjusts the number of servers as appropriate.

Formation like shipping lanes for applications pipelines

individual deployment units are called Cells.

Mesos-based DC/OS (Data Center Operting System)

No-script deploy?

The main difference is that developers don’t need to learn Jenkins and Chef.

References

https://www.oreilly.com/ideas/using-a-single-codebase-for-your-cloud-native-app

More on DevOps

This is one of a series on DevOps:

Wilson Mar