devops - the talent dividend

27
DevOps - The Talent Dividend through a Brain & Behavioural Looking Glass Stefano Picozzi [email protected]

Upload: stefano-picozzi

Post on 12-Apr-2017

222 views

Category:

Technology


0 download

TRANSCRIPT

DevOps - The Talent Dividend through a Brain & Behavioural Looking Glass

Stefano Picozzi [email protected]

Architects set high level guidelines and principles.

Developers develop with operational sensibilities.

InfaOps deliver application centric infrastructure.

Libertarian Paternalism

DevOps[Portmanteau]Collabora0onbasedonteaming,feedbackandexperimenta0on

ATROPHY Describes a loss of neurons and the connections between them.

PLASTICITY Brain’s ability to reorganize itself by forming new neural connections.

Complex Adaptive Organization

The Hard Problems In Software Delivery

operate deploy release test build code plan why

Adapted from: http://www.continuousautomation.com/wp-content/uploads/2014/08/solution-s-curve.png

Lean

DevOps

Continuous Delivery

Continuous Integration Agile Development

Collaboration

Valu

e

Existential Value Delivery

FIRST MILE LAST MILE

Topics

PEOPLE

PROCESS TECHNOLOGY

DEVOPS

SEMAT

MATURITY MODEL

SCARF

MOTIVATION CONTAINERS

AUTOMATION

People

The SCARF Model for Collaboration Away Toward

Threat Reward

Status

Certainty

Autonomy

Relatedness

Fairness

Perception of control over environment. A sense of choice dramatically reduces stress.

Deciding whether others are ‘in’ or ‘out’ of a social group. Empathy response is more active within the tribe or with friends.

Brain is a pattern matching machine. More resources are expended processing novel situations.

Significant determinant of health and longevity. Challenge to status can generate strong threat response.

Equitable outcomes. Sense of unfairness triggers strong threat response.

SCARF on DevOps Away Toward

Threat Reward

Status

Certainty

Autonomy

Relatedness

Fairness

Small, long-lived, cross-skilled team

Embed Developers into Operations

Embed Operators into Development

A success exemplar

Shared decision making

Skills transfer to new model

Relevancy via enablement

Equity in remuneration

KPIs and MBO alignment

Career growth outside management

Measurement time horizons

Executive sponsorship

Leadership authenticity

Reinforcement of behaviours

Role transition clarity

Motivation 3.0

AUTONOMY

PURPOSE

MASTERY

The urge to get better at something that matters.

The desire to do something that has meaning and is important.

To be self-directed. Control increases compliance but autonomy increases engagement.

Motivation 3.0 for DevOps

AUTONOMY

PURPOSE

MASTERY

Self-service

Automation

Service catalogs

Immediacy

Frictionless

Choice

Roles based access

Enable, get out of the way

Accessibility

From Novice to Expert

Multiple form factors

Scale invariance

Tooling range

Training and enablement

Open innovation

Community engagement

Collaboration

CI/CD integration

Image provenance

Stateful applications

Security and scalability

Logging

Telemetry

Middleware …

W W S D

Technology

Make the desirable choice easier. Reward adherence.

The ABCs of DevOps

Teaming Feedback Experimentation

Standardization Automation Autonomy

Increased velocity Improved quality Reduced waste

A B CNegative

Positive Reinforcement

Containers Lets first agree on a standard work product

Includes runtime dependencies

Easy to deploy

Portable across host systems

Isolates applications on O/S

Encourage microservices

Immutable

Multitenancy

Illusion of infinite capacity

Containers-as-Code

An amphora is a type of container of a characteristic shape and size, descending from at least as early as the Neolithic Period. Amphorae were used in vast numbers for the transport and storage of various products, both liquid and dry, but mostly for wine.

Orchestration Then let's agree on a uniform QoS strategy

●  Open source project conceived by Google

●  Google and Red Hat are the top contributors

●  Fast paced and highly collaborative

○  ~150-200 commits/week

●  Provides ○  Container orchestration on multiple hosts ○  Container grouping ○  Self-healing ○  Service discovery

Flock of seagulls as metaphor for self-organizing swarm. Also name of an 80s new wave English synthpop band known for the song "I ran (So Far Away)".

Dante’s Automation

Bare Metal

Virtualization

Containerization

O/S

IaaS

PaaS

FaaS

A journey of abstraction

Explore/Exploit

Cloud agnostic deployment strategies

Automation Tool Chain The Siren's song of Automation

Complex IaaS specific

scripting

CI/CD IaaS CI/CD PaaS

•  Infrastructure as code •  Infrastructure specific •  Specialized skills to operate •  Technical automation debt •  Non-transferable hosting model

•  Containers as code •  Infrastructure agnostic •  Accessible for all skill-sets •  Supported platform •  Transferable on/off premise

Making Technology Accessible

Develop Build Deploy

DevOps

ITS … THE VIBE plus a little bit of process

DevOps Maturity Model 1 (Initial) 2 3 (Improved) 4 5 (Optimizing)

Culture & Organization

Teams organised based on platform/technology. Defined and documented processes. Low cooperation or power-oriented. ⬤

Agile Adoption One backlog per team. Adopt agile methodologies. Remove team boundaries. Share the pain. Modest cooperation or rule-oriented.

Scaled Agile Approach (SAFe or other) Extended team collaboration. Remove boundary dev/ops. Common process for all changes.

Cross-team continuous improvement. Teams responsible all the way to production.

Cross functional teams. High cooperation or performance-oriented.

Test & Verification

Automated unit tests on every build. Separate test environment. ⬤

Some automatic integration tests. Code analysis. Test coverage analysis.

Automatic component tests. Full automatic integration tests. Behaviour-driven development.

Full automatic acceptance tests. Manual exploratory testing. Automatic performance tests. Automatic security tests.

Verify expected business value. Defects found and fixed immediately (roll forward).

Information & Reporting

Baseline process metrics. Manual reporting. ⬤

Measure the process. Static code analysis. Automatic reporting.

Automatic generation of release notes. Pipeline traceability. Reporting history.

Report trend analysis. Real time graphs on deployment pipeline metrics.

Dynamic self-service of information. Customizable dashboards.

Build & Deploy Centralized version control. Automated scripts for building software. Nightly builds. No management of artifacts. Manual deployment. ⬤

Continuous Integration Polling or triggered builds (commit hook). Fail builds if they do not compile and pass unit tests. Any build can be re-created from source control. Management of build artifacts. Builds are not left broken.

Deployment Pipelines Fail builds if more general quality is not met. Automated provisioning of environments. Automated deployments. Standardized environment templates. Standard deployment process for all environments.

Team prioritises keeping codebase deployable over doing new work. Orchestrated deployments. Blue/green deployments.

Zero touch continuous deployments.

Data Management Data migrations unversioned and performed manually. ⬤

Automated and versioned changes to datastores.

Changes to datastores automatically performed as part of the deployment process.

Automatic datastore changes and rollbacks tested with every deployment.

Release Infrequent and unreliable releases. ⬤

Painful infrequent but reliable releases. ⬤

Infrequent but fully automated and reliable releases in any environment.

Frequent fully automated releases. Deployment disconnected from release.

No rollbacks, always roll forward.

Team is roughly at this stage already and is not targeting this area for next phase.

Team is targeting these activities as critical for next phase.

Team is targeting these activities as potential work for next phase.

Cells without status indicate stages the team has not achieved fully and are not being focused on for the next phase.

READY

SET

GO

The current state is assessed, target end state agreed to, and then appropriate intervention is designed and applied.

It is easier to act yourself into a new way of thinking, than it is to think yourself into a new way of acting - Millard Fuller

Insights & Summary

About one second before insight solutions, there is an increase in alpha-band (10-12Hz) activity measured over posterior occipital-parietal cortex/

Solvingthehardproblemofvaluedelivery-Butconnectini+a+vewithexisten+aldriver

Wearefixingabehaviouralproblem-Premiseiscausa+onofteamingwithoutcomes

CultoftheChangeAgent-ButcheckthatagendanothijackedbyTechnora+

Automa0onleadingtoself-serviceisac0onable-ButbewaretheSiren'ssongofautoma+on

Mindsetofenablementandadherence-Notcontrolandcompliance

Startwithsmallempoweredgroup-Thenpromotesuccessforviralenablement

https://www.openshift.com/promotions/kubernetes.html https://www.openshift.com/

promotions/docker-security.html

https://t.co/4KH6iSZZ2H https://www.openshift.com/promotions/for-developers.html

DevOps - The Talent Dividend through a Brain & Behavioural Looking Glass

Stefano Picozzi [email protected]

https://blog.emergile.com/2016/06/21/openshift/