IIBA DevOps- Is the Marriage working? By Shiboo Varughese, EVP, - - PowerPoint PPT Presentation

IIBA – DevOps- Is the Marriage working?

By Shiboo Varughese, EVP, Cirrus Labs Shiboo@cirruslabs.io

CirrusLabs Services

We custom-build technology solutions for your business, because we understand that solutions aren’t one-size-fits-all. Take a look inside our proven solutions process that carries

• ver into all areas of our technical expertise:
• End-to-End Solutions
• Integration & Automation
• Ongoing Support
• Training & Education

CLOUD SERVICES. Our cloud experts help you on your cloud journey from start to finish. We perform assessments, assist in vendor selection, support your team through coaching & training, execute your migration, provide managed services, and help keep your cloud secure. INTERNET OF THINGS. Our holistic view of the IoT Ecosystem delivers the right integrated solution for your needs. Our IoT assessment helps you discover new business opportunities. We deliver quality solution design & development, and our device integration support makes running your IoT solution feel effortless.

• AUTOMATION. Our speed-to-market analysts help you adopt leading

automation processes. We’ll assess your current practices using a DevOps scorecard, enable end-to-end DevOps, support your team through coaching & training, help you pick & configure the right tools, and implement robotic process automation.

• ANALYTICS. Our data analysts & scientists operate as an

extension of your business analytics team, providing key insights to inform your business decisions. Our data lab approach enables real-time analytics and leverages machine learning to help you get the most out of your data.

Passion & Realization Cooperation & Reunion

01 03

Conflict & Rebellion Commitment & Maturity

04 02 03 06

DevOps –Phases of Marriage

Lifespan of a Fortune 500 Company

Motivations

Comparing the Fortune 500 companies in 1955 to the Fortune 500 in 2017, there are only 61 companies that appear in both lists.

Source: Steve Denning and Mark Perry

Half a century ago, the life expectancy of a firm in the Fortune 500 was around 75 years. Now it’s less than 15 years and declining even further.

Metrics from High Performing Orgs

Company Deploy Frequency Deploy Lead Time Reliability Customer Responsiveness Amazon 23,000/day Minutes High High Google 5,500/day Minutes High High Netflix 500/day Minutes High High Facebook 1/day Hours High High Twitter 3/week Hours High High Most Enterprises Once every 9 months Months or Quarters Low/Med Low/Med

Source: The Phoenix Project: A Novel About IT, DevOps and Helping Your Business Win

What’s DevOps worth to ANY organization?

“DevOps is instrumental in reducing software deployment cycles from days to minutes. We are able to deploy new features quickly and safely, and rapidly respond to market changes.”

Rob Cornish CTO, International Securities Exchange

“Organizations that have implemented DevOps practices are up to five times more likely to be high-performing than those that have not …”

PuppetLabs

“High-performing organizations deploy code 30 times faster with 50 percent fewer failures.”

PuppetLabs 2013 State of DevOps Report

“… a 17 to 23 percent improvement in business in the form of increased revenue, faster time-to-market, improved competitive positioning and enhanced customer experience.”

CA Technologies Press Release , Sept 12, 2013

How Organizations Achieve High Performance

HIGH Performers

• 89% are using infrastructure version control
• 82% are using automated code deployments

Source: PuppetLabs 2014 State of Devops:

Passion and Realization : Business

Case for DevOps.

The Steps Are Always The Same

Recognize Need Define Requirements Build & Test Reap Benefits Deploy & Operate

… TO SAVE M ONEY HERE

People are now more expensive and they’re not very good at pre-specifying requirements

Over time the economics have changed

Recognize Need Define Requirements Build & Test Reap Benefits Deploy & Operate

COLLABORATE HERE …

Prototyping Rapid Application Development Iterative / Agile / XP

… TO SAVE M ONEY HERE

Developing fast doesn’t matter if you can’t deliver fast

But some things remain painful

Deploy & Operate

Deployment still hurts.

• Configuration issues
• Human error
• Risky: Testing was probably compromised

We’re only willing to suffer deployments monthly or quarterly so velocity drops

ODDS ARE HIGH THAT YOU’RE DELIVERING LATE

Recognize Need Define Requirements Build & Test Solution Deploy & Operate Solution Reap Benefits

4-6 Weeks

To deliver a simple change

3-4 Weeks

Average time to isolate a defect

YOU MAY BE BUILDING THE WRONG THING

Recognize Need Define Requirements Build & Test Solution Deploy & Operate Solution Reap Benefits

20%

Of total defect volume is in the requirements

TESTING USUALLY TAKES PLACE VERY LATE

Typical QA tests must wait until there is some working application so that the tests can be “recorded” and then played back later. But if you have to wait, you’re already slowing down your delivery

YOU’RE PROBABLY TRYING TO FIX IT WITH TESTING

a National Research Council study on “dependable software” concluded that testing is no longer sufficient to ensure an application will be reliable, efficient, secure, and maintainable. When all sources and kinds of software defects are considered as a set, coding defects only comprise about 17% of the total defect volume. Testing in all its forms seldom tops 85% in total defect removal

• efficiency. A combination of pretest removal and testing removal is

necessary to top 95%... As it happens, 95% is a good demarcation line between good quality and mediocre quality.

You probably have quality problems

Recognize Need Define Requirements Build & Test Deploy & Operate Reap Benefits

41%

Delays in integration, configuration and testing of apps

51%

Apps rolled back because of defects leaking into production

45%

Troubleshooting and fine-tuning delays in production

Waterfall SDLC

Traditional Application Development Lifecycle - Fragmented

Recognize Need Define Requirements Build & Test Reap Benefits Deploy & Operate

Business IT Development IT Operations Business Business

Initiate Plan Design Develop Deploy Test

Less Collaborative Teams

You always screw up my deployments. Ops team does not know what they are doing..

Developer

No, you may not deploy that software

Infosec Operations

We can’t keep up with your insane demands! I have to keep my env up and running with 99.5 %SLA

Users

This is so not what we needed.

Rebellion: The complications

Not Defining DevOps.. DevOps is NOT… … Another Name for Agile … NoOps … A Job Title … (Just) Culture … (Just) Process … (Just) Tools

Traditional Key Success Factor Modern Thinking Command & Control Management Style Autonomous Conservative Attitude to Change Experimental Silo Organization Structure Collaborative Project – Focus Delivery Focus Product-Centric Waterfall Delivery Model Iterative Large(Huge) Batch Size Smallest Monolothic Systems Arch Loosely Coupled Proprietary Technology Open Sourced Manual Process Automated

Organizational Success Factors

Waterfall Iterative Agile With DevOps

Cooperation & Reunion:

(Re)Organizing the value Chain

The three ways of DevOps

The Three Ways: The Principles Underpinning DevOps. Gene Kim.

• Best candidate systems are designed for change and
• architected from the beginning to take advantage
• f DevOps definition aspects:

– Lean – Automated – Measurable

DevOps is not a “Big Bang”: start small & prove value

Gartner, September 2014

What do your customers want? Results!!

Rapid Delivery

• f Business

Value

It’s time to re-think the value chain

You must excel in four areas

Rapid Delivery of Business Value Early Problem Detection Rapid Repair of Problem s Rapid Developm ent Agreed-Upon Business Need and Success M etrics

Once you know what you want and how you’re going to measure it then you must be able to build it fast and be able to see and fix problems before they bother your customers.

For now, let’s set aside the need and metrics; that’s a whole presentation unto itself.

Strong skills and experience are required

Rapid Repair

• f Problem s

Early Problem Detection Rapid Developm ent

Skilled, Experienced Team

Agreed-Upon Business Need and Success M etrics

Collaboration Tools (Some) Common Engineering Tools

To go fast, all of the engineering capabilities—initial development and then rapid identification and repair of problems—require a more strongly skilled and properly experienced team.

They must be able to communicate and solve problems well. They need support with good engineering and collaboration tools. Rapid Delivery of Business Value

Don’t guess. Know. Test early and often

Rapid Delivery of Business Value

Continuous Testing

Early Problem Detection Rapid Repair of Problem s Rapid Developm ent Skilled, Experienced Team

Instrumentation & Operational Monitoring

Agreed-Upon Business Need and Success M etrics Collaboration Tools (Som e) Com m on Engineering Tools

If you want to deliver results fast, then testing cannot wait.

You Can’t Test It if it Hasn’t Been Delivered Somewhere

Continuous testing needs continuous delivery

Rapid Delivery of Business Value Continuous Testing

Continuous Delivery Continuous Integration

Early Problem Detection Rapid Repair of Problem s Rapid Developm ent Skilled, Experienced Team Instrum entation & Operational M onitoring Agreed-Upon Business Need and Success M etrics Collaboration Tools (Som e) Com m on Engineering Tools

… and a great side-effect of “continuous” behavior is that you’re also testing your deployment and operating

• procedures. By the time you

get to a production release, it’s almost a non-event

This works better with elastic compute

Rapid Delivery of Business Value Continuous Testing Continuous Delivery Continuous Integration Early Problem Detection Rapid Repair of Problem s Rapid Developm ent Skilled, Experienced Team Instrum entation & Operational M onitoring Agreed-Upon Business Need and Success M etrics Collaboration Tools (Som e) Com m on Engineering Tools

Elastic Compute Infrastructure Infrastructure Automation

THE STATE OF DEVOPS: The Value Chain Map

Value Evolution Genesis Custom Product / Rental Commodity / Utility

Rapid Delivery

• f Business

Value Continuous Testing Continuous Delivery Infrastructure Autom ation Continuous Integration Early Problem Detection Rapid Repair of Problem s Rapid Developm ent Skilled, Experienced Team Instrum entation & Operational M onitoring Agreed-Upon Business Need and Success M etrics Collaboration Tools (Som e) Com m on Engineering Tools Elastic Com pute Platform

Issues:

• Speed of delivery

depends in part on minimizing the duration

• f the feedback cycle to

the developer

• Without supporting

continuous techniques around testing, delivery and integration, that cycle takes a long time

• Longer cycles also include

more changes which lengthens the repair time and increases the risk of change

• Latent defects are more

expensive to fix

Application Release Cost ( An Example)

Efficiency Factors Labor Hours TOTAL Average Number of Engineers to Do a Deployment 5 engineers Average time taken per deployment 4 hours Average total time spent on release deployments 5 engineers x 4 hours = 20 hours Fully loaded cost of an engineer $70 per hour Average total cost per deployment 20 hours x $70 per hour = $1400 Number of customers in the environment (assume 50% of clients have quarterly releases) 100 As is quarterly deployments per year [ $1400 x 100 x 4] $560,000

§ Assumptions below are based on environment assuming 100 customers § Current Annual Costs= $560,000 Manual Deployment Story: 1 Distributed App, approx.20 components in 50-60 servers. Week End Deployment : All Hands on Deck. Major Effort. Exhaustion for All. Finally Finished. Executives Notified and Relieved that everything is completed. Until ???

Application Release Cost ( An Example)

Efficiency Factors Labor Hours TOTAL Average Number of Engineers to Do a Deployment 5 engineers Average time taken per deployment 4 hours Average total time spent on each release deployments 5 engineers x 4 hours = 20 hours Fully loaded cost of an engineer $70 per hour Average total cost per deployment 20 hours x $70 per hour = $1400 Number of customers in the environment (assume 50% of clients have quarterly releases) 100 As is quarterly deployments per year [ $1400 x 100 x 4] $560,000 New average time for deployment with 1 engineer 30 minutes. or .50 hrs. New Total cost for deployment (0.5 x 70 x 100 x 4) $14,000

§ Assumptions below are based on environment assuming 100 customers § Current Annual Costs= $560,000 reduced to $14,000 Push Button Deployment 1 Distributed App, approx.20 components in 50-60 servers. Week End Deployment : All Automated with easily understandable metrics.

Commitment & Maturity:

The trends

DevOps DataOps DevSecOps

01 03 02

Business/Product Owner defines & Prioritizes Product Backlog Team Collaborates To Complete Code Review. Tags the developer for comments/updates. Development pulls tasks from Sprint Board, details design using confluence based design

• templates. Tags team members

for Design review. Once design is reviewed, development starts for the tasks. Code is tagged for Continuous Build. Automated tests are run. SONAR results are reviewed. Release package built Team does Sprint Planning based on user stories & prioritizes into sprint. Team members do estimates & refines the Sprint backlog. Business Stake Holder/Product Owner captures Requirements & maps into User Stories. Product Owner refines the Product Back

• Log. Project Team Plans for Sprint

based delivery based on backlog. The pre built artifacts with the features are delivered to environments using Continuous Deployment. Development completes the development tasks & assigns for review. Adds a ticket in Jira for Individual Code Review. Commits the code into the repo (triggers continuous build process) referring the sprint task id for better traceability.

A B C D E F G Continuous Delivery Model

Agile + DevOps Continuous Delivery

Measure maturity by metrics

• MTTR
• Deployment Frequency
• Lead Change time
• Change Failure Rate

The emergence of DevOps introduced new set of practices and tools – which improved the quality, predictability, scale and speed of software development and deployment. These learnings of DevOps can be applied to DataOps. At the core, it is about integration of the different facets of the organization.

Source: Carnegie M ellon University

DevSecOps

Security is no longer an after thought, but well Integrated with software development

Conclusion

“It is not the strongest of the species that survive, nor the most intelligent, but the one most responsive to change.” – Charles Darwin

“Fail quick, fail often, recover quickly.”

A DevOps marriage can still work if

• Understand the real passion of merging Ops and Dev
• Move through the Rebellions
• Consider a step by step approach which make sense to the team
• Create a Mature model and continually refine it..

Thank You

Shiboo Varughese

• Developer, Technologist and Architect
• Currently EVP for CirrusLabs ( A Startup minded product and

Services company in Atlanta - http://cirruslabs.io)

• Twenty years of driving innovative solutions & business
• utcomes to both IT & the Business
• Previously Senior Enterprise Architect & leader of Daugherty’s

Software Architecture & Engineering and Information Management Line of Service

• Former Director and Chief Architect for Enterprise solutions for

Xerox. My email : Shiboo@cirruslabs.io