Introduction to DevOps

DevOps
Presented by:
Ahmed Adel El-Adl

What is DevOps ?
2
◇ DevOps = Dev ( Development ) + Ops ( Operations )
◇ Wikipedia Def:
◇ “DevOps is a software engineering Culture and Practice that aims at unifying
software development (Dev) and software operation (Ops).
◇ DevOps aims at shorter development cycles increased deployment frequency and more
dependable releases in close alignment with business objectives”

◇ DevOps is
◇ DevOps is first a Culture of
collaboration between developers
and operations people
◇ This culture has given rise to a set
of Practices
◇ DevOps is a grassroots movement,
by practitioners, for practitioners
3
◇ DevOps is NOT
◇ DevOps is NOT tools, but Tools are
essential to success in DevOps
◇ DevOps is NOT a standard
◇ DevOps is NOT a product
◇ DevOps is NOT a job title

This Session Will
Cover…
◇ DevOps Culture – The culture of collaboration between Dev and Ops.
◇ DevOps Practices – The practices which support the goals of DevOps culture.
◇ DevOps Tools – The tools that help implement DevOps practices.
4

A Brief History of DevOps
◇ DevOps grew out of the Agile software development movement.
◇ Agile seeks to develop software in small, frequent cycles in order to deliver
functionality to customers quickly and quickly respond to changing
business goals.
◇ DevOps and Agile often go hand-in-hand.
5

Timeline of DevOps
6
◇ 2007: Agile software development was gaining popularity, but it was also suffering
from a growing divide between development and operations.
◇ 2008: Patrick Debois and Andrew Shafer met at the Agile2008 Conference in
Toronto, Canada. They began to start conversations and seek others interested in
bridging the divide between dev and ops.
◇ June 23, 2009: John Allspaw and Paul Hammond gave a talk at Velocity
Conference: “10+ Deploys Per Day: Dev and Ops Cooperation at Flickr.”
Patrick was watching via livestream. People began discussing it via twitter.
◇ October 30 2009: Patrick hosted the first DevOpsDays in Ghent, Belgium; a
conference for both devs and ops. The conversation continued on Twitter: #devops
◇ DevOps grew into an organic, grassroots movement all over the world and spawned
many tools to support the practices valued by DevOps.

Today ……..
◇ The DevOps movement has not stopped growing since 2009, and is no longer a small,
niche movement.
◇ It is now a mainstream, spawned a large variety of tools, completely changed the IT
industry forever.
7

DevOps Culture
◇ DevOps culture is about collaboration between Dev and Ops.
8
◇ Under the traditional separation
between Dev and Ops, Dev and
Ops have different and opposing
goals.
◇ With DevOps, Dev and Ops work
together and share the same goals.

A Story of
DevOps vs. Traditional Silos
9

Traditional Silos
◇ Devs write code
◇ “throw it over the wall” to QA
◇ Code bounces back and forth between Dev and QA as QA discovers problems
and Devs fix them
◇ Finally, it is ready for production
10

Traditional Silos
◇ QA/Dev “throws the code over the wall” to Operations
◇ Oh no! There’s a problem. Ops throws it back over the wall to Dev
◇ Each group’s domain is a “black box” to the other groups
◇ “Our systems are fine; it’s your code!”
◇ “But the code works on my machine!”
11

Traditional Silos
What Went Wrong?
◇ Dev and Ops are black boxes to each other, which leads to finger pointing:
￭ Because Ops is a black box, Devs don’t really trust them.
￭ And Ops doesn’t really trust Dev.
◇ Dev and Ops have different priorities, which pits them against each other:
￭ Ops views Devs as breaking stability.
￭ Devs see ops as an obstacle to delivering their code.
◇ Even if they WANT to work together:
￭ Dev is measured by delivering features, which means deploying changes.
￭ Ops is measured by uptime, but changes are bad for stability.
12

Downsides of
Traditional Silos
◇ “Black boxes” lead to finger pointing.
◇ Lengthy process means slow time-to-market.
◇ Lack of automation means things like builds and deployments are
inconsistent.
◇ It takes a long time to identify and fix problems.
13

With DevOps
◇ Devs write code
◇ Code commit triggers automated build, integration, and tests
◇ QA can get their hands on it almost immediately
◇ Once it is ready, kick off an automated deployment to production
14
Build → Integrate → test → Deploy

With DevOps
◇ Oh no! The latest deployment broke something in production!
◇ Fortunately, automated monitoring notified the team immediately
◇ The team does a rollback by deploying the previous working version, fixing the
problem quickly
◇ An hour later, the dev team was able to deploy a fixed version of the new code
15
Build → Integrate → test → Deploy Monitor

DevOps
What Went Right?
◇ Dev and Ops worked together to build a robust way of changing code quickly and reliably:
￭ Both Dev and Ops worked together to prioritize both speed of delivery and stability
◇ Automation led to consistency:
￭ Building, testing, and deploying happened the same way every time
￭ Building, testing, and deploying happened much more quickly and more often
◇ Good monitoring, plus the swift deployment process, ensured problems could be fixed even
before users noticed them:
￭ Dev and Ops worked together up front to build good processes
￭ Even though a code change caused a problem, users experienced little or no downtime
16

1- Build Automation
◇ Build automation: automation of the process of preparing code for deployment to a live
environment.
◇ Depending on what languages are used, code needs to be compiled, linted, minified,
transformed, unit tested, etc.
◇ Build automation means taking these steps and doing them in a consistent, automated way
using a script or tool.
◇ The tools of build automation often differ depending on what programming languages and
frameworks are used, but they have one thing in common: automation!
18

What does build automation look like?
◇ Usually, build automation looks like running a command-line tool that builds code using
configuration files and/or scripts that are treated as part of the source code.
◇ Build automation is independent of an IDE.
◇ Even if you can build within the IDE, it should be able to work the same way outside of
the IDE.
◇ As much as possible, build automation should be agnostic of the configuration of the
machine that it is built on.
◇ Your code should be able to build on someone else’s machine the same way it builds on
yours.
19

Why do build automation?
◇ Fast : Automation handles tasks that would otherwise need to be done manually.
◇ Consistent : The build happens the same way every time, removing problems and
confusion that can happen with manual builds.
◇ Repeatable : The build can be done multiple times with the same result. Any version of
the source code can always be transformed into deployable code in a consistent way.
◇ Portable : The build can be done the same way on any machine. Anyone on the team
can build on their machine, as well as on a shared build server. Building code doesn’t
depend on specific people or machines.
◇ Reliable : There will be fewer problems caused by bad builds.
20

2- Continuous Integration(CI)
◇ Continuous Integration (CI): the practice of frequently merging code changes done
by developers.
◇ Traditionally, developers would work separately, perhaps for weeks at a time, and
then merge all of their work together at the end in one large effort.
◇ Continuous integration means merging constantly throughout the day, usually
with the execution of automated tests to detect any problems caused by the
merge.
◇ Merging all the time could be a lot of work, so to avoid that it should be
automated!
21

What does Continuous Integration look like?
◇ Continuous integration is usually done with the help of a CI server.
◇ When a developer commits a code change, the CI server sees the change and
automatically, performs a build, also executing automated tests.
◇ This occurs multiple times a day.
◇ If there is any problem with the build, the CI server immediately and automatically
notifies the developers.
◇ If anyone commits code that “breaks the build” they are responsible for fixing the
problem or rolling back their changes immediately so that other developers can continue
working.
22

Why do Continuous Integration?
◇ Early detection of bugs : If code doesn’t compile or an automated test fails, the
developers are notified and can fix it immediately. The sooner these bugs are
detected, the easier they are to fix!
◇ Eliminate the scramble to integrate just before a big release : The code is constantly
merged, so there is no need to do a big merge at the end.
◇ Makes frequent releases possible : Code is always in a state that can be deployed to
production.
◇ Makes continuous testing possible : Since the code can always be run, QA testers
can get their hands on it all throughout the development process, not just at the end.
◇ Encourages good coding practices : Frequent commits encourages simple, modular
code.
23

3- Continuous Delivery &
Continuous Deployment
◇ Continuous Delivery (CD): the practice of continuously maintaining code in a
deployable state.
◇ Regardless of whether or not the decision is made to deploy, the code is always in
a state that is able to be deployed.
◇ Some use the terms continuous delivery and continuous deployment
interchangeably, or simply use the abbreviation CD.
◇ Continuous Deployment: the practice of frequently deploying small code changes
to production.
◇ Continuous delivery is keeping the code in a deployable state. Continuous
deployment is actually doing the deployment frequently.
◇ With continuous deployment, deployments to production are routine and
commonplace. They are not a big, scary event
24

What does Continuous Delivery and Continuous
Deployment look like?
◇ Each version of the code goes through a series of stages such as automated build,
automated testing, and manual acceptance testing. The result of this process is an artifact
or package that is able to be deployed.
◇ When the decision is made to deploy, the deployment is automated.
◇ If a deployment causes a problem, it is quickly and reliably rolled back using an automated
process(hopefully before a customer even notices the problem!).
◇ Rollbacks aren’t a big deal because the developers can redeploy a fixed version as soon as
they have one available.
◇ No one grips their desk in fear during a deployment, even if the deployment does cause a
problem.
25

Why do Continuous Delivery and
Continuous Deployment?
◇ Faster time-to-market : Get features into the hands of customers more quickly
rather than waiting for a lengthy deployment process that doesn’t happen often.
◇ Fewer problems caused by the deployment process : Since the deployment
process is frequently used, any problems with the process are more easily
discovered.
◇ Lower risk : The more changes are deployed at once, the higher the risk. Frequent
deployments of only a few changes are less risky.
◇ Reliable rollbacks : Robust automation means rollbacks are a reliable way to
ensure stability for customers, and rollbacks don’t hurt developers because they
can roll forward with a fix as soon as they have one.
◇ Fearless deployments : Robust automation plus the ability to rollback quickly
means deployments are commonplace, everyday events rather than big, scary
events.
26

4- Configuration Management
◇ Configuration Management: maintaining and changing the state of pieces of
infrastructure in a consistent, maintainable, and stable way.
◇ Changes always need to happen – configuration management is about doing them in a
maintainable way.
◇ Configuration management allows you to minimize configuration drift – the small
changes that accumulate over time and make systems different from one another and
harder to manage.
27

What does configuration management
look like?
◇ Here is an Example:
◇ You need to upgrade a software package on a bunch of servers:
￮ Without good configuration management, you log into each server and perform
the upgrade. However, this can lead to a lot of problems. Perhaps one server was
missed due to poor documentation, or perhaps something doesn’t work while the
versions are temporarily mismatched between servers, causing a lot of downtime
while you do the upgrade.
￮ With good configuration management, you define the new version of the software
package in a configuration file or tool and automatically roll out the change to all
of the servers.
◇ Configuration management is about managing your configuration somewhere
outside of the servers themselves
28

Why do configuration management?
◇ Save time : It takes less time to change the configuration.
◇ Insight : With good configuration management, you can know about the state of all
pieces of a large and complex infrastructure.
◇ Maintainability : A more maintainable infrastructure is easier to change in a stable
way.
◇ Less configuration drift : It is easier to keep a standard configuration across a
multitude of hosts.
29

5- Orchestration
◇ Orchestration: automation that supports processes and workflows, such as
provisioning resources.
◇ With orchestration, managing a complex infrastructure is less like being a builder and
more like conducting an orchestra.
◇ Instead of going out and creating a piece of infrastructure, the conductor simply
signals what needs to be done and the orchestra performs it:
￭ The conductor does not need to control every detail.
￭ The musicians (automation) are able to perform their piece with only a little bit of
guidance.
30

What does Orchestration look like?
◇ Here is an example:
￭ A customer requests more resources for a web service that is about to see a heavy
increase in usage due to a planned marketing effort.
￭ Instead of manually standing up new nodes, operations engineers use an
orchestration tool to request five more nodes to support the service.
￭ A few minutes later, the tool has five new nodes are up and running.
◇ A much cooler example:
￭ A monitoring tool detects an increased load on the service
￭ An orchestration tool responds to this by spinning up additional resources to handle the load.
￭ When the load decreases again, the tool spins the additional resources back down,
freeing them up to be used by something else.
￭ All of this happens while the engineer is getting coffee
31

Why do Orchestration?
◇ Scalability : Resources can be quickly increased or decreased to meet changing needs.
◇ Stability : Automation tools can automatically respond to fix problems before users see
them.
◇ Save time : Certain tasks and workflows can be automated, freeing up engineers’ time.
◇ Self-service : Orchestration can be used to offer resources to customers in a self-service
fashion.
◇ Granular insight into resource usage : Orchestration tools give greater insight into how
many resources are being used by what software, services, or customers.
32

6- Monitoring
◇ Monitoring: The collection and presentation of data about the performance and
stability of services and infrastructure.
◇ Monitoring tools collect data over things such as:
￮ Usage of memory
￮ cpu
￮ disk i/o
￮ Other resources over time
￮ Application logs
￮ Network traffic
￮ etc.
◇ The collected data is presented in various forms, such as charts and graphs, or in the
form of real time notifications about problems
33

What does Monitoring look like?
◇ Real-time notifications:
￭ Performance on the website is beginning to slow down
￭ A monitoring tool detects that response times are growing
￭ An administrator is immediately notified and is able to intervene before downtime
occurs.
◇ Analysis:
￭ Something went wrong in production last night.
￭ It’s working now, but we don’t know what caused it.
￭ Luckily, monitoring tools collected a lot of data during the outage.
￭ With that data, developers and operations engineers are able to determine the root
cause (a poorly performing SQL query) and fix it.
34

Why do Monitoring?
◇ Fast recovery : The sooner a problem is detected, the sooner it can be fixed. You want to
know bout a problem before your customer does!
◇ Better root cause analysis : The more data you have, the easier it is to determine the
root cause of a problem.
◇ Visibility across teams :Good monitoring tools give useful data to both developers and
operations people about the performance of code in production.
◇ Automated response :Monitoring data can be used alongside orchestration to provide
automated responses to events, such as automated recovery from failures.
35

7- Microservices
◇ Microservices: A microservice architecture breaks an application up into a collection of
small, loosely-coupled services.
◇ Traditionally, apps used a monolithic architecture. In a monolithic architecture, all
features and services are part of one large application.
◇ Microservices are small: each microservice implements only a small piece of an
application’s overall functionality.
◇ Microservices are loosely coupled: Different microservices interact with each other
using stable and well-defined APIs. This means that they are independent of one
another ( Microservices vs Monolith )
36

What does Microservices look like?
◇ There are many different ways to structure and organize a microservice architecture
◇ For example, a pet shop application might have:
￮ A pet inventory service
￮ A customer details service
￮ An authentication service
￮ A pet adoption request service
￮ A payment processing service
◇ Each of these is its own codebase and a separate running process (or processes). They can
all be built, deployed, and scaled separately
37

Why do Microservices?
◇ Modularity : Microservices encourage modularity. In monolithic apps, individual pieces
become tightly coupled, and complexity grows. Eventually, it’s very hard to change
anything without breaking something.
◇ Technological flexibility : You don’t need to use the same languages and technologies
for every part of the app. You can use the best tool for each job.
◇ Optimized scalability : You can scale individual parts of the app based upon resource
usage and load. With a monolith, you have to scale up the entire application, even if
only one aspect of the service actually needs to be scaled.
◇ Microservices aren’t always the best choice. For smaller, simpler apps a monolith might
be easier to manage.
38

Introduction to
DevOps Tools
39

The Role of Tools in DevOps
◇ DevOps is NOT a set of tools.
◇ But how can we achieve high speed of delivery while maintaining stability? TOOLS!.
◇ The DevOps community has created a wide range of powerful tools.
◇ Part of doing DevOps is identifying the tools you need and learning how to use them.
40

1- Build Automation Tools
◇ Build automation – Automated processing of code in preparation for deployment.
◇ What tools you use for build automation usually depend on programming languages
and frameworks
◇ A few examples:
￮ Java – ant, maven, gradle
￮ Javascript – npm, grunt, gulp
￮ Packer – build machine images and containers
41

2- Continuous Integration Tools
◇ Continuous Integration – Continuously merging code into a single branch or mainline
◇ CI tools usually consist of a CI server that integrates with source control
◇ When source code is changed, the server responds by executing an automated build
◇ A few examples:
￮ Jenkins – Open source ,Widely used ,Java based
￮ TravisCI – Open source , Built around Github integration, Executes builds in clean VMs
￮ Bamboo – Enterprise product by Atlassian, integration with other Atlassian products like JIRA,
BitBucket and Confluence
42

3- Configuration Management Tools
◇ Configuration Management – Managing and changing the state of pieces of infrastructure in
a consistent and maintainable way
◇ A few examples:
￮ Ansible – Open source , Declarative configuration (YAML), No control server needed.
￮ Puppet – Declarative configuration, Manage state through a UI, Pushes changes to clients
using a control server and agents installed on clients.
43

4- Tools for Virtualization and
Containerization
◇ Virtualization : Managing resources by creating virtual rather than physical machines
◇ Examples:
￮ VMWare ESX and ESXi
￮ Microsoft Hyper-V
￮ Citrix XenServer
◇ Containers: Lightweight, isolated packages containing everything needed to run a software
◇ Require fewer resources than VMs, VMs contain an entire OS plus virtual versions of hardware
◇ Containers have the bare minimum needed to run the software
◇ Docker – Docker is currently the leading container technology
◇ Containers are still relatively new but very useful for DevOps!
44

5- Tools for Monitoring
◇ Monitoring – Collecting and presenting data about the state and performance of applications
◇ There are different types of monitoring:
￭ Infrastructure monitoring (ISM) : focuses on things related to infrastructure like CPU, ram
￭ Application Performance Monitoring (APM) : focuses on performance and stability of
individual parts of an like response times, logs
45
ISM APM ISM + APM

6- Tools for Orchestration
◇ Orchestration – automation that supports processes and workflows, such as provisioning resources
◇ Lets you do things like:
￭ Scale up and scale down applications on request.
￭ Auto scale applications based on usage.
￭ Create self-healing systems by spinning down unhealthy nodes and replacing them with new ones.
◇ A few examples:
￮ Docker Swarm: – Docker-native, Orchestration for Docker containers
￮ Kubernetes:– Open source, Orchestration server, Manage containerized apps across multiple
hosts
46

Next Steps…..
◇ Tool Quick Starts:
￮ Docker (+Docker Certified Associate)
￮ Ansible
￮ Kubernetes
￮ Git
￮ Jenkins
￮ Puppet
￮ Serverless Concepts
48
◇ Cloud Platforms:
￮ Google Cloud Platform
￮ Azure
￮ Amazon Web Services

Introduction to DevOps

Recommended

More Related Content

What's hot (20)

Similar to Introduction to DevOps (20)

Recently uploaded (20)

Introduction to DevOps