12 factor app - Core Guidelines To Cloud Ready Solutions

2
Yeah, this is ME ...
About ME (The Author) >
Technology Evangelist | ML Enthusiast | Solution Designer & Architect
LinkedIn Profile: www.linkedin.com/in/ksiddiqui
MS CS from Lahore University of Management & Sciences(LUMS), 2004
Service Delivery Manager @ Nisum Tech, 2018
I am a tech enthusiast, started my career in 2003 as a plain software engineer, and now after 16 yrs
working as a solution architect for the last 6-7 years, currently working in NISUM Tech in Karachi. Besides
that I have taught various courses from Operating Systems to Network Security, and engaged in research
work while doing my post graduate.
You can also find me at:
Slideshare @ https://www.slideshare.net/kashifalisiddiqui
Quora @ https://www.quora.com/profile/Kashif-Ali-Siddiqui
Pinterest @ https://www.pinterest.com/kuantum9/
Medium @ https://medium.com/@kashif.ali.siddiqui

Ü Set of principals
Ü A collection of guidelines
Ü Methodologies
Ü Software engineering practices
That helps one in ...
• Designing a perfect software architecture for large scale/volume use
• Running a fluent engineering process for complex ever changing business
workflows
• Making sure the timely release of new features, changes, fixes
• Maintaining a software solution over time
3

https://12factor.net/
4
The main Internet resource online on the
topic with all the primary details.

• The methodology was drafted by developers at Heroku
• Was first presented by Adam Wiggins circa 2011
• The principals were laid down as foundation to the Heroku platform, but later
due to their generic and platform independent implementation they are
released as fundamental guidelines for any cloud ready application
5
-- Heroku --
• Heroku is a cloud platform as a service (PaaS) supporting several programming languages, starting from
Ruby, and now Java, Node.js, Scala, Clojure, Python, PHP, and Go
• A polyglot platform, allowing developers to build, run and scale applications in a similar manner across most
languages
• Heroku was acquired by Salesforce.com in 2010
https://www.heroku.com/

6
Easy maintenance, and
reusability
Using declarative formats for setup automation
Maximize portability between
environment
Using clean contract with the underlying operating
system
Cloud ready architecture Fulfilling all the cloud based solution requirements
Maximize automation By minimize divergence between development and
production
Provides scaling and
resilience
Using distributed system's design patterns
Platform independent
support
Provide identical rules for all kind of software tech
stacks, languages

• Will most probably end up
• Designing a rigid software ...
• With too much dependencies on the environment and setup
• No room for future extensions
• No or little flexibility towards third party integrations
• Creating a monolithic software
• Lossing capability to scale for large volume access and/or data processing
• Lossing resilience to various failures
• Decreasing software delivery efficiency and efficacy
• Creating an inefficient engineering process and/or its execution
• Etc ...
7

• Ofcourse software developers ... developing
• SAAS (Software As A Service) solutions
• Distributed software solutions
• Cloud application solutions
• Online Internet based services
• Online games / gaming services
• Social networking services
• Internet application support services
• Enterprise solutions
• On-prem solutions
• Large scale, and/or highly resilient solutions
8

9
I Codebase
One codebase tracked in revision control, many deploys
VII Port Binding
Export services via port binding
II Dependencies
Explicitly declare and isolate dependencies
VIII Concurrency
Export services via port binding
III Config
Store config in the environment
IX Disposibility
Maximize robustness with fast startup and graceful shutdown
IV Backing Services
Treat backing services as attached resources
X Dev/Prod Parity
Keep development, staging, and production as similar as possible
V Build, Release, Run
Strictly separate build and run stages
XI Logs
Treat logs as event streams
VI Processes
Execute the app as one or more stateless processes
XII Admin Processes
Run admin/management tasks as one-off processes

10
Definitions ...
Codebase Ü The source code for the application that will
be compile or interpreted before execution.
Repository (repo) Ü Codebase managed within a VCS
(Version Control System) system.
Deploy Ü An installation of the app with a specific
configuration for a specific use case and/or audience.
Do's
• An app codebase must be stored in a repository managed by
a VCS such as Git
• Must be 1-to-1 correlation between the codebase and the
app
• Means, a distributed app can have multiple codebases,
one for each distributed module
• Fact: A codebase can trigger multiple deploys ... where a
deploy is a running instance of app ... deploys shares code but
at different versions
Don'ts
• Multiple apps cannot share code ... such code must be
factored out as shared libraries
• Hence there will be no separate codebase for each deploy
I Codebase One codebase tracked in revision control, many
deploys

11
II Dependencies Explicitly declare and isolate dependencies
Definitions ...
Dependency Ü Any library module used by the app which
is either developed/managed by the developer or by a 3rd
party vendor requires. A dependency is identified by its
name, version, vendor, and download source.
Dependency Manifest Ü A document within the codebase
that lists all the dependencies for the app. E.g. pom.xml,
build.gradle, gemfile, etc.
Dependency Manager Ü A separate module which is part
of the build toolchain that processes the dependency
manifest, manages and download all the required
dependencies from their sources before the codebase is
taken to the compilation. E.g. Maven, Gradle, PIP, etc.
Dependency Isolation Ü A separate module which is part
of the build toolchain which provides dependencies
isolation for each app within an environment such that a
dependency within an isolation (virtual environment)
cannot be used by another app from the other isolation.
E.g. Virtualenv, etc.
Do's
• The software codebase should declares all of its
dependencies, along with their specific versions (if required)
in a dependency declaration manifest
• The software codebase during build process should uses
dependency manager, aka packaging system, to fetch all
required dependencies from their sources and maintain them
in a local repository
• A dependency isolation tool is used to prevent accidental
import of unwanted dependencies or their other versions
from the shared environment
• The full and explicit dependency specification is applied
uniformly to both production and development.
• Fact: In Java world, Maven, Gradle are the dependency
managers
Don'ts
• A software codebase does not relies on the implicit existence
of dependencies in the build environment

12
III Config Store config in the environment
Definitions ...
Config Ü A collection of parameters that hold values that
are used by the app's codebase in 1) making strategic
decisions, 2) integration with backing services, 3)
communication, 4) processing requests, 5) formating
outputs, etc. An app’s config is everything that is likely to
vary between deploys (staging, production, developer
environments, etc).
Evironment Variable Ü A collection of key-value pairs
defined in the environment. A process receives these
variables from it's parent process (creator). Hence these
variables are not being fetched by the process, but rather
handed over from it's parent that get those from it's
parent and so on. These variables can hold all sort of data
like information of the OS, environment, dependencies,
and configuration for the process itself.
Do's
• An app requires strict separation of config from its code
• Config can also be made available from a file separately define in the
system registry like /etc away from the app, but it is not a generic way
of passing config, as it is not possible in other non-Linux
environments like Windows
• An app should get config from the environment as enviroment
variables as it is much more platform independent way
• Various app packaging and containerization runtimes, and
orchestration systems provides facility to define config for the app
thru environment variables based on the deploy type
Don'ts
• An app should not store config within code files, or as separate files
within codebase
• An app should not retrive config with tech stack and/or language
specific APIs
• An app should not use config groups like production, staging, qa, dev
config as it make it quite rigid and prune to various issues

13
IV Backing Services Treat backing services as attached resources
Definitions ...
Resource Ü A helper acquired thru request or created by
the app to perform certain task.
Backing service Ü A backing service is any resource the
app consumes over the network as part of its normal
operation; e.g. datastores, messaging/queueing systems ,
SMTP services, and caching systems. A backing service can
be managed by the app developer e.g. database, or by
some third party e.g. email gateway. A backing service is
pretty much different from an embedded resource, as with
the later one, one can have assumption regarding
availability and resilience. A backing service can be
attached and detached at it's own will or due to other
uncontrollable circumtances.
Do's
• The app should always assume backing services as resources whom
life cycle management is the responsibility of the app itself
• If an app process utilizes the services of another process, then the
other process becomes the backing service for the first one
• The code for an app should make no distinction between local and
third party services ie resources managed by dev vs managed by
others
• The service access parameters must be part of the app's config so
that each deploy can have a choice to attached to other service of
same type
• An app should have logic and functionality to handle each cases
regarding the resource attachment and detachment scenarios
Don'ts
• An app must not assume a resource (either local or third party)
as always available, and/or resilient to failures

14
V Build, Release, Run Strictly separate build and run stages
Definitions ...
Build Ü The build process takes the code from the repo, acquire
all the dependencies from their sources, and generates an app
build that can be use for release. The process can take code thru
compilation and optimization processes.
Release Ü The release process takes the app build and based on
the deploy type acquire the necessary app config and generates a
release build, that will be ready for execution in its deploy
environment.
Run Ü It is a process of actually executing the app, and creating
its process(es).
BUILD = codebase + dependencies + assets
RELEASE = BUILD + config
RUN = run process against RELEASE
Do's
• Fact: An app codebase transform into a deploy thru three steps; 1)
build, 2) release, and 3) run
• Hence all the stages must be taken as separate processes that can be
run on their convinent times; “build” when the developer pushes new
changes, “release” when deploy is required, and “run” when the app
is required to become available
• The “run” can run multiple times with the same “release” on the
environment, “release” can run multiple times with same “build” for
different deploy type configs, and “build” can run multiple times
• The “build” process will use its own ID to tag build artifacts, whereas
“release” process will either use its own ID or add suffix to the build
ID
• The build and release artifacts should be stored separately for future
access and reference
Don'ts
• An app must not have a unified process that can build, create release,
and does the run on the target
• A CI pipeline must not implements the continous delivery and continous
deployment phases.

15
VI Processes Execute the app as one or more stateless processes
Definitions ...
Process Ü The running of a release in a deploy environment.
App State Management ... The app is designed such that,
• if it's process crashed, terminated abruptly, there won't be any
loss of state or data except the processing of the current
requests.
• if app's multiple instances are be launch in parallel, they will
not need to be aware of each other
• there will be no synchronization requirements between it's
process instances.
• any instance can be brought down any time without any need
to backup any state or service data.
With Stateful Design ...
• The app processes reduces the overall app resilience.
• Increases the process liability.
• Makes the process unexpendable, reduces desposibility.
• Requires to have state syncing and consistency assurance
procedures.
• Reduces the process overall efficiency and efficacy.
• Makes an app harder to upgrade in deployment environments.
Do's
• Fact: An app is executed in its deployment environment as one or a
collection of processes
• A business workflow must be designed as a process instead as a
thread
• An app process(es) must be designed to run statelessly and share
nothing ... Any state that requires persistent must be handled by the
backing services (e.g. database)
• An app process should identify the incoming request's session or
context rather than assuming the previous one
• Should save the sessions in a database rather than holding it in its
memory
• Uses single transaction only caches
Don'ts
• An app's process not necessarily get all the subsequent requests from
the same session/context
• An app's process must not rely on sticky session feature
• An app's process cannot assume itself as fault free, and thus can get
terminated on number of conditions other than the code itself

16
VII Port Binding Export services via port binding
Definitions ...
Port Ü A virtual interface for the TCP/UDP protocol on
which the app communicates with the other app.
Port binding Ü A procedure in which an app process
associates its communication interface with a TCP port,
and other processes then establish comm link to the first
process on its binded port.
Do's
• An app should be fully self-contained; means it does not requires
runtime injection of a web server/container
• An app should only bind to a TCP/UDP port rather than the complete
address set i.e. IP address and TCP/UDP port.
• An app port binding should be configurable; not hard coded in the
codebase
• Fact: Apps with port binding will bring flexibility of getting run within
the same environment provided that all the other processes are
binded to different unique ports
• An app thru dependency declaration, should add an embedded web
server/container to the release bundle that gets port binding config
from the environment
• Fact: Apps with port binding and embedded web server/container
enables to have a distributed architecture of multiple processes
acting as backing services to others
Don'ts
• An app's process must not tie up itself with the IP address along with
port

17
VIII Concurrency Scale out via the process model
Definitions ...
Concurrency Ü Ability to run similar tasks in parallel to
load balance, and to scale up the solution capability to
take on more load or data volume for processing.
Concurrency can be achieve either at process level or at
thread level.
Thread Ü A thread is a unit of execution within a process
that can run independently and in parallel. A process can
have from one to N number of threads, all running in
parallel, and can share and access each other memory and
resources.
Process level concurrency Ü A type of concurrency where
the task parallelism is achieve by running two or more
processes, and each process executes the tasks.
Thread level concurrency Ü A type of concurrency where
the task is executed in a thread, and multiple threads are
spawned to parallelize tasks execution within a single
process boundry.
Do's
• An app should run their business workflow at the process level; not at
the thread level
• The app should use thread level concurrency for specific smaller tasks
like communication, accessing data or resources, or crunching
numbers or data into information
• Hence the main business workflow concurrency is achieved by
instantiating more processes on same or separate environments, and
within the business workflow various parallelism among the subtasks
is acheived thru threads
• Fact: Concurrency thru process enables isolation of one business
workflow from others; gives failure and starvation protection to each
business process from others; provides an effective and efficient way
of scaling load, and maximizing resilience towards failures
• An app should be configured to manage its output streams, crash
recovery, auto-start policy on system restart/reboot
Don'ts
• An app process neither should deamonize nor create PID files; but
rather behave like normal processes that can be controlled from the
central process manager like systemd in Linux

18
IX Disposibility Maximize robustness with fast startup and graceful
shutdown
Definitions ...
Disposibility Ü An app's process characteristic that makes
it expandable which further enables the process manager
and system orchestrator to start and teardown these app's
processes on demand without going into various issues of
data/state synchronization, resource integration, etc.
An app ...
→ with stateless design,
→ providing business workflow execution at process level,
→ and that takes configuration from the environment
are more elastic in nature and thus can be easily
terminated and restarted immediately without considering
their previous runtime state.
Do's
• An app process should be design in a way that it can be tear down,
terminated, and restart again in a moment's notice
• An app process should have minimum startup time
• An app process should shut down gracefully on terminate signal
• An app process should also be robust against sudden failure, and
should be architectured to handle unexpected, non-graceful
termination without lossing the inprogress workload requests
• Fact: An app with the above characteristics enables fast elastic scaling,
rapid deployment of code and config changes, and robustness of
production deploys
• For a worker process, the jobs taken from the queue should have a
state thru which the process can mark a job in progress, and later
delete the job or mark it done when it is completed; whereas in case
of a job in progress state for more than a predefine time, can be
retried by other worker process
Don'ts
• Most of the usual resources like memory, open files, pipes, FIFOs are automatically released when a process terminates in any
condition. However there are resources that requires an explicit release procedure like TCP sockets. In today's world, the
frameworks usually take care of these resources, but if the app is handling the sockets, then it should take care of its graceful
shutdown.

19
X Dev / Prod Parity Keep development, staging, and production as
similar as possible
Definitions ...
Dev -> Prod parity Ü An app usually have gaps between development and
production due to the following reasons:
• Development completion period ... Time taken by the developers to
complete a feature, a change or a fix, and have it reviewed and pushed
to the codebase.
• Release deployment ... Time taken by the devops / CD to deploy the
new build on production site that often include testing, reviewing, etc.
• Environment dissimilarities ... The changes between the dev and
production environment like changes in tech stack, servers, backing
services, or their versions, etc.
Continous integration Ü A process that automates the task of creating
builds, validating the build thru unit tests, and automated functional testing.
Continous delivery Ü A process that automates the task of creating
releases for production enviroments, validating the release thru automated
integration and UAT testing on staging environment.
Continous delivery Ü A process that automates the task of deploying the
tested release on production environment with a strategy the favors
smooth transition from old version to the newer one.
Do's
• An app and its engineer process should be design to favor continous
integration, continous delivery, and continous deployment
• An app's engineering process should minimize the dev and
production gap to the maximum by
• Making features/changes/fixes small so that they can be
developed in less time, and requires lesser time in testing
• Making release deployment more efficient and effective so that
new builds with new features/changes/fixes can be deployed
quickly on the target environments
• Trying to make dev and production environment as similar as
possible by using infrastructure that can spawn or create
environments from a template on will to create any required
environments like dev, QA, staging, etc.
• Use mock layers for the backing services for an apps to do functional
testing
Don'ts
• Donot allow local deploys for integration testing; always use
spawnable environments

20
XI Logs Treat logs as event streams
Definitions ...
Logs Ü Logs are the explicitly printed/written statements from the
app's code to document start and end of events and situations.
These statements are outputted to a single resource in a uniform
structure/pattern in textual format; usually one statement per line.
The output resource is hoped to keep these log statements for an
enough longer period where they can be further process and
reviewed.
Loging Ü A processing of outputting log statement to the logging
device/resource.
Event Ü Events are somethings that happens at a given time or
place representing some case, consequence, or an effect.
Streams Ü A flow of data units of same structure and/or size that
undergoes same/identical processing.
Logs as event streams Ü A stream of logs statements that
represents chronologically sorted events in the processes. Each
statement must be uniform in structure.
Do's
• Fact: Logs provide visibility into the behavior of a running app; have
no fixed beginning or end, but flow continuously as long as the app is
operating
• Fact: Logs are the stream of aggregated, time-ordered events
collected from the output streams of all running processes and
backing services
• Fact: Logs in their raw form are typically a text format with one event
per line
• An app should write its logs to its output stream
• An app output stream should be configurable from the environment
so as to redirect logs as per requirement
• An app's log must follow a definite pattern with one line per log entry
• An app's log should be created as they will be pushed to an analyzer
thru an aggregator for further analysis, and event processing
Don'ts
• An app should never concern with managing its logging stream
• An app should never output its logs to files

21
XII Admin Processes Run admin/management tasks as one-off processes
Definitions ...
One-off process Ü A kind of process that terminates when it is done with
its specific task. Means it doesnot runs forever. It runs separately from the
actual app, either by the user having permission, and/or by the system on
certain events (as scheduled job).
Admin processes Ü Admin processes are one-off processes that have no
direct relation to any business workflow within the app, means they donot
perform any app's business responsibility, but rather are executed to
support them or make them more efficient. E.g. processes for data
migration (import/export), database cleaning, process maintenance, etc.
Benefits of having admin processes ...
• They have very little code, only for the task at hand.
• They can be executed anytime and in any order when there is a need.
• There execution is isolated from the app's own processes, hence any
failure either in admin process or app's own process donot impact
other; and similarly either of them cannot starve other, neither there
will be any security issues.
• Since admin processes are separate, they can by-pass various business
logic and validations on data access for speedier processing.
Do's
• Fact: An app often comes with various one-off administrative
processes for housecleaning, and maintenance tasks like cleaning
temporary/used/deleted/malformed data, etc
• Fact: These one-off admin processes are run within the production
environment where the actual apps deploys and runs using the app's
config
• The one-off admin processes should be implemented as separate
processes from app's own processes
• Each one-off admin process should have its own separate app
• The one-off admin processes apps can be bundled with the actual
app bundle, but they should be kept separately from the normal app,
and their executable access should also be limited to fewer users
Don'ts
• Hence a one-off admin process should not be made part of the actual
app that can be design to run on certain time and situation

23
Deploy env templates Ü that can be use to clone new
environments
Env spawner Ü that can create new env from the
templates by cloning them and applying declared
transformations
4 deploy envs Ü which are production (blowup),
staging, QA and dev, all cloned by the spawner from
their respective templates
App A and App B Ü are actual business apps with an
embedded listner (queue consumer, web container
respectively) binded to specific ports M and N
respectively
App A and App B Configs Ü are the externalized
configurations provided from outside and injected into
App A and B as env variables at deploy time before
execution
One-off Admin Processes A and B Ü are the two
admin apps separate from the actual business apps A
and B, working with the actual backing services
Backing services Ü are divided into two; managed,
and externally managed, are those managed by the
dev and by the 3rd party respectively. Here business
App B is also a backing service to App A.

24
Deploy environment Ü is the required infrastructure for app deployment with (production) configuration.
App A and App B Ü are the actual business apps that implements certain business concerns or workflows.
App A and App B Configs Ü are the configs for the respective business apps A and B provided separately
with the environment based on the environment type.
One-off Admin Process A Ü Is one of the one-off admin app run separately from the business apps but
shipped with the same build bundle and utilizes the same backing services, as well as the business app
config.
Apps processes Ü The business apps based on the single responsbility principal (SRP) can further breaks
into a set of processes where each process implements a single responsibility. Here App A is compose of
two processes, whereas App B is compose of one process. An admin app can also mapped to one or more
processes. Here it is mapped to one process.
Concurrency Ü The parallelism in the processing of multiple business workflows is achieved by running
two or more processes within the app based on the load requirement, and capability. Similarly admin app
can have two or more process instances based on the load requirements. These process instances can be
run on same or different computing node.
Statelessness & state management Ü is one of the core design element in the business apps. Since an app
is design to be stateless, all its process instances are then stateless. Hence due to that the env manager can
tear down an instance or start a new one based on the requirement without worrying about the state
management. If the process working with the state (as of App A), the changed state is immediately pushed
to the backing service (database) that keeps the states for all the instances. Hence with this rudimentary
form of state management, the instances get free from the state synchronization and state consistency
management.
Logging Ü is done by all the process instances of each App A and B, and these are aggregated into a single
stream and reroute to an admin console and/or to an analysis software.
Backing services Ü are divided into two; managed, and externally managed, are those managed by the dev
and by the 3rd party respectively. Here business App B is also a backing service to App A.

25
1) An engineering process/strategy that favors ...
Ü cleaner and leaner codebase utilizing repository capabilities to seggregate deploy type code
Ü quicker delivery to deploys; every deploy is like hot-fix
Ü quicker fixes to bugs and issues
Ü enhanced testing quality by enabling devs to test on production similar environments
3) A software architecture that supports ...
Ü extensibility in apps
Ü single responsibility apps
Ü statelessness
Ü concurrency thru scaling processes
horizontally
Ü tech agnostic logging that can be relayed to
any other system
Ü high disposibility; the ability to start and
shutdown immediately
2) A software deployment strategy that enables ...
Ü easy setup of deploy environments thru the use
of spawnable environments
Ü heterogeneity in deployment environments
Ü similar or identical dev and production
environments
Ü flexibility in deployment strategy
Ü efficient configuration injection
Ü isolation between actual and admin processes

26
# Design Patterns
1 Automation Enforced thru separate build, release, run processes, and minimized dev-prod parity. This includes ...
• Continous integration, continous testing, continous delivery, continous deployment
2 Software
architecture
Enforced thru architecture characteristic requirements like ...
• Software compartmentization
• Responsibility segregation
• Stateless, self-contained processes
3 Deployment
patterns
Enforced thru the use of spawnable environments that means using app containers or virtual machines
along with env orchestration solutions
4 Configuration
externalization
Enforced to make configuration external to the app and providing it from external source mainly as
environment variables
5 Communication
resilience patterns
Enforced to handle all failure cases with backing services by implementing patterns such as:
• Retry, timeout, fallback, circuit breaker, fail fast, isolation, bulkheads, location transparency
7 Service tracing and
log aggregation
Enforced thru logging to process output stream and letting the environment to route the stream to
external processors (analyzers)

27
Definition
Ü A container being a standard unit of software, packages up code and all its
dependencies along with the required configuration so the application runs
quickly and reliably from one computing environment to another.
• Hence a container is just an environment for the target application that contains the
application, its runtime, its dependent libraries, configurations, and other resources.
• A container usually shares OS and its system libraries with the host environment.
• In case of having containers on a different OS, then emulation is being used. For
instance setting up Linux based containers on Windows OS.
• An application in a container with the matching host OS gives almost the same
performance as it is deployed directly on the host environment.
• An containerized application can be serialized in an image, and then the image can be
replicated and deployed on other hosts as per need.
• Hence a service containerized can be saved as an image, and then can be used
later for repeated number of deployments.
• Containers provides isolation to the application such that
• an application cannot see others in the other containers
• an application cannot impact / affect others applicaitons
• an application cannot access resources in other application containers
• an application can have dedicated runtime and dependencies for specific version
For example ...
• Docker containers
• CoreOS Rkt
(Rocket)
• Mesos containerizer
• Etc
Docker Ü is an application container software that not only provides the
capability of a containerization, but also management. Docker is developed
primarily for Linux, where it uses the resource isolation features of the Linux
kernel such as cgroups and kernel namespaces, and a union-capable file
system such as OverlayFS and others to allow independent "containers" to
run within a single Linux instance, avoiding the overhead of starting and
maintaining virtual machines (VMs).
Application containers help implements some of the
principals of 12 factors ... which are:
• Config → by providing a mechanism to define env
variables as part of the container definition that
becomes available to the containerized app when
launched.
• Backing service → by providing a strategy to deploy all
internal backing services as containerized apps so that
they can be redeploy again and again.
• Processes → by providing process manager thru which
each app process can be managed externally.
• Port binding → by providing port mapping facility thru
which the containerized app ports can be mapped to
other ports.
• Dev/Prod parity → by providing a mechanism to create
images of builds and then spawn containers from them
on need basis. Hence easing the process of creating
environments.
• Logs → by providing standard output stream collection
facility; hence all the standard output can be routed to
external processors.

28
Definition
Ü Container orchestration is all about managing the lifecycles of
containers, especially in large, dynamic environments with multiple
hosts/nodes.
Software teams use container orchestration to control and automate many tasks:
• Provisioning and deployment of containers
• Defining communication paths
• Defining resilience schemes
• Designing security parameter
• Redundancy and availability of containers
• Scaling up or removing containers to spread application load evenly across
host infrastructure
• Movement of containers from one host to another if there is a shortage of
resources in a host, or if a host dies
• Allocation of resources between containers
• External exposure of services running in a container with the outside world
• Load balancing of service discovery between containers
• Health monitoring of containers and hosts
• Configuration of an application in relation to the containers running it
For example ...
• Kubernetes
• Docker Swarm
• Apache Mesos
• (and Marathon)
Container orchestrators help implements most of the principals of 12
factors. Since they manages containers and their images, hence all the
principals enforced by containers (discussed previously) are also being
enforced by them indirectly. The principals they support are:
• Backing Services → by providing functionality to define backing
services, and their availability requirements that includes how
scalable and resilient they should be. Moreover using advance
features one can easily define integrations to these backing
services. And reusing them among two or more environments.
• Build, Release, Run → by providing support for various strategies
for deploying, and executing containerized apps.
• Processes → by providing app definition thru which ops can
declare apps, and how their processes will be created as containers
and how they will be managed.
• Concurrency → by providing controls to define process
instantiation, scalability, and auto restart capabilities.
• Disposibility → by providing support for process startups,
shutdown, and their performance measurements.
• Dev/Prod Parity → by providing infrastructure to define
deployments, and controls to manage them as per need.
• Logs → thru functions to capture logs from standard output
streams and route them to admin consoles.

29
Definition
Ü Platform as a Service (PaaS) or Application Platform as a Service (aPaaS) is a
category of cloud computing services that provides a platform with environments set
up with app runtime allowing customers to develop, run, and manage applications
without the complexity of building and maintaining the infrastructure typically
associated with developing and launching an app.
The advantages of PaaS are primarily that it allows for higher-level programming with dramatically
reduced complexity; the overall development of the application can be more effective, as it has built-
in/self up-and-down ramping infrastructure resources; and maintenance and enhancement of the
application is thus easier.
Disadvantages of various PaaS providers as cited by their users include increased pricing at larger
scales, lack of operational features reduced control, and the difficulties of traffic routing systems.
PaaS solutions usually build on top of container
orchestration solutions, and with that they indirectly
provides support for all the 12 factors supported by
container orchestration solutions discussed previously.
PaaS solutions provides their own CLI based toolchain thru which
devs and ops can further impl. and/or improve 12 factors which are:
• Codebase → by integrating code repo with the app deploys and
hence making continous deployment an easy process,
manageable from the PaaS CLI client.
• Dependencies → by handing over the dependency management
to PaaS.
• Config → by binding config for an app with PaaS own
configuration using CLI.
• Build, Release, Run → by providing a simple easy way for
creating build and release artifacts, and then doing the deploy
and run using the PaaS CLI.
• Disposibility → by providing an app framework that make sure
of the disposibility characteristics for an app.
• Dev/Prod Parity → by providing support thru CLI for deploy of
app on various environment types with least parity.

30
Vendor Service URL
Google Cloud Platform (GCP) App Engine https://cloud.google.com/appengine/
Amazon's AWS Elastic Beanstalk https://aws.amazon.com/elasticbeanstalk/
Microsoft's Azure App Services https://azure.microsoft.com/en-us/services/app-service/
IBM Bluemix (Cloud Foundry) https://www.engineyard.com/
VMware Pivotal Cloud Foundry https://www.vmware.com/solutions/photon-pcf.html
SAP HANA Cloud Platform https://cloudplatform.sap.com/index.html
Redhat Openshift https://www.redhat.com/en/technologies/cloud-computing/openshift
Orcal Cloud Platform (OCP) Oracle Cloud PaaS https://cloud.oracle.com/en_US/tryit
Saleforce Force.com https://www.salesforce.com/products/platform/overview/
Pivotal Pivotal Web Services https://run.pivotal.io/
Heroku Heroku Dynos https://www.heroku.com/dynos
Software AG LongJump https://www.crunchbase.com/organization/longjump
CenturyLink AppFrog https://www.ctl.io/
Engine Yard Engine Yard https://www.engineyard.com/
Platform.sh Platform.sh https://platform.sh/
Taken from lists:
https://www.techradar.com/news/best-paas-provider
https://www.cbronline.com/cloud/10-of-the-best-paas-providers-4545381/
https://stackify.com/top-paas-providers/

31
Definition
Ü Serverless computing is a cloud-computing execution model in which the cloud
provider runs the server, and dynamically manages the allocation of machine
resources based on the incoming computation requests. Pricing is based on the
actual amount of resources consumed by an application, rather than on pre-
purchased units of capacity. It can be a form of utility computing.
It is one step above PaaS (discuss previously) where the infrastructure is created and starts execution
from the time of app deployment, whereas in serverless solutions the infrastructure is only
spawned/used when there is a computation request/need. And hence there is often a limitation to
terminate the computation within a predefined timeout otherwise the app execution is halted by the
platform.
Serverless computing can simplify the process of deploying code into production. Scaling, capacity
planning and maintenance operations may be hidden from the developer or operator. Serverless code
can be used in conjunction with code deployed in traditional styles, such as microservices. Alternatively,
applications can be written to be purely serverless and use no provisioned servers at all.
Difference from PaaS solutions Ü The major difference is how the infrastructure is provisioned. In
PaaS, the infrastructure (VMs or containers) is provisioned when the app is deployed, whereas in
serverless deployment, it is provisioned when the app is requested for computation. Hence based on
this, the billing in PaaS started from the time of app deployment, whereas in serverless the billing is
done when the app is computing on a request. Moreover the infrastructure in PaaS remains intact
through out the app deployment timeperiod, whereas in the case of serverless deployment the
infrastructure is tear down once the app computation is completed/ended/terminated.
The other difference is on the resource limitations aspect. In PaaS, usually there is no limitation to the
app other than memory and/or storage (sometimes), whereas in serverless there is a lot of limitation on
CPU cycles, memory utilization and storage. Hence the app in serverless is expected to complete its
processing in a definite time period utilizing not more than declared memory with very little or no
harddisk storage.
Some serverless platforms
→ Google's GCP - Cloud Functions (https://cloud.google.com/functions/)
→ Amazon's AWS - Lambda Functions (https://aws.amazon.com/lambda/)
→ Microsoft's Azure - Functions (https://azure.microsoft.com/en-us/services/functions/)
These serverless platform's enables developers to code their apps as discreate, small
modules; each one is self contained, and can be executed independently, and can fetch and
save data/state from/to a database. Cloud platform often also provide orchestration tools
to integrate these serverless apps to various other apps and events.
Programming languages Ü Java, JavaScript, Python, GoLang, Ruby, etc.
Serverless solutions provides same supports for 12 Factor apps as
PaaS solutions and in some principals exceeds such as:
• Concurrency → by parallelizing app's module instead of the
whole app, and as per need bases.
• Disposibility → by futher breaking the app into self-contained
app-functions (modules) that can start and terminate
independently on need bases.
• Dev/Prod Parity → by making deployments more simple, and
lightweight with ease of provisioning based on demand.
• Admin Processes → by enabling admin tasks to be
implemented as app-functions that can run on separate
provisioned context when there is a need.

32
Type Softwares/Libraries/Frameworks 12 Factor App Support
VCS Git, CVS, SVN, etc Codebase Use to store app's codebase, and provide management for version
control.
Dependency managers
+ Dependency isolators
Maven, Graddle, PIP, VirtualEnv, etc Dependency Use to manage app's dependencies by keeping all required
dependencies and updating them as their new versions get available.
Continuous integration Jenkins, TeamCity, etc Build-Release-Run Use to host and run pipelines that are responsble to create builds and
releases for certain deploys.
Continuous
deployment
Spinnaker, XL Deploy, DeployBot, AWS
CodeDeploy,
Build-Release-Run, Dev/Prod Parity Use to deploy releases on target environments.
Containers, Container
orchestrators
Docker, Rkt, Kubernetes, etc Build-Release-Run, Processes,
Concurrency, Dev/Prod Parity, Port
Binding
Use to build environment infrastructure for the app's processes.
Embedded
servers/containers
Tomcat, Jetty, Flask, Nameko, etc Processes, Concurrency, Port Binding Use to contain an app, and expose its interface over some predefined
port and protocol.
Cloud, PaaS CLI clients gcloud, aws, azure-cli, pf, etc Codebase, Config, Process,
Concurrency, Build-Release-Run, Logs
Use to create, push, pull, start, stop an app deployment on the
infrastructure.
Scripting languages Bash, Python, Groovy, Powershell,etc Build-Release-Run, Dev/Prod Parity,
Admin Processes
Use to provide aid in CI/CD pipelines, and to develop admin apps.
Configuration
management
Chef, Puppet, Ansible, etc Config, Build-Release-Run, Dev/Prod
Parity
Use to define environment configuration and automate the
configuration process through out the deploy environments.
Database systems Redis, MongoDB, Cassandra, MySQL,
Kafka, RabbitMQ, Apache Camel, etc
Backing Services, Processes Use to hold app's state, and provide other data related facilities.

>Cloud ready solution
You can develop applications without getting into hassel of explicitly handling cases and requirements for cloud deployment.
>Ensures scalability and resilience to the design
You will get enormous scalability and resilience capabilities without actually doing any explicit development for that.
>Promotes distributed software solution design patterns
You will automatically get compartmentization in your app with distributed modules implementing distributed design
patterns.
>Ease in system maintenance
You will get several advantages in system maintenance by utilizing various tools and softwares that provides automation in
testing, delivery, deployment, monitoring, and reporting, making software maintenance an easy job.
33

>Enables easy and effective automation
You will transform your automation jobs into most effective processes possible that makes efficient deloyments a reality.
>Streamline engineering processes
Your engineer process will become more efficient in making speedier deliveries and more responsive to the change
requirements without compromizing the build qualities.
>Increases software quality and delivery efficiency
You will be able to increase your app's build quality due to more efficient and effective test automation, app
compartmentization, no/minimum dev-prod parity, and quicker delivery mechanism.
>Helps avoid the cost of software erosion
You will be able to make quicker and periodic maintenance on your app that includes adding new features, changes and fixes,
hence preventing your app from lossing customers, and keeping their interest alive in your software solution.
34

For more information on microservice architecture, and how 12
factors are being used there, please refer to my another
presentation @
35
Understanding MicroSERVICE
Architecture with Java Spring-Boot

12 factor app - Core Guidelines To Cloud Ready Solutions

More Related Content

What's hot (20)

Similar to 12 factor app - Core Guidelines To Cloud Ready Solutions (20)

Recently uploaded (20)

12 factor app - Core Guidelines To Cloud Ready Solutions