EDB® WHITE PAPER

Deploying
Postgres
Databases in
Containers

GETTING STARTED

EnterpriseDB | www.enterprisedb.com
 STNETNOC
03 Introduction

04 What Is Driving Containers?

07 Why Containerized Databases Now?

12 Who Benefits From Using Containers?

14 Conclusion

Published April 2019

EnterpriseDB, EDB and EDB Postgres are trademarks of EnterpriseDB Corporation.
Other names may be trademarks of their respective owners. Copyright© 2019. All rights reserved. 20190405 WWW.ENTERPRISEDB.COM
Introduction
Today it is safe to say that containers and container orchestration have left the hype
phase and are quickly becoming mainstream application development technologies.
The container growth trends are compelling:

According to a survey of 600+ IT decision makers conducted by ClearPath

Strategies for the Cloud Foundry Foundation, 30% of enterprises have already
deployed containers and 42% are in the evaluation stage.
From a Sysdig report we learn that the median number of containers per server
now stands at 15, up 50% over the previous year.
Containers are not only being experimented with in dev/test environments. A
Portworx survey of more than 500 IT professionals revealed that in 2018, 83% of
the enterprises running containers were using them in production, versus 67% in
2017.
Diamanti’s Container Adoption Benchmark Survey clearly shows that container
use cases are becoming far more diverse, including new cloud-native applications
(~55%), lightweight stateless apps (~39%), modernizing legacy apps (~31%), and
databases (~30%).
Last but not least, 451 Research reports that revenues from container
technologies in 2018 were $1.53 billion, which is a little more than three times the
revenues in 2015. By 2020, container technology revenues are expected to grow
by another 75%, reaching $2.7 billion.

But what is a container, exactly? A container is a resource-isolated software object that

packages all the code, configurations, and dependencies needed to execute and run an
application on any Linux or Windows platform, regardless of the environment (on-
premises, public or private cloud, or even bare metal). Because multiple containers can
share a single operating system (OS), they are significantly more lightweight than virtual
machines (VMs). Container orchestration platforms, such as Kubernetes, automatically
launch clusters of containers and manage their entire lifecycle, making it easier to
deploy containers at scale.

In this white paper, we explore at some depth what is driving the growth of containers, as
well as the opportunities and challenges of a specific container use case: databases.

EDB® WHITE PAPER / PAGE 3 WWW.ENTERPRISEDB.COM

What Is
With more and more born-in-the-cloud
apps, containers, along with

Driving microservices and serverless, are an

integral part of the trend towards

Containers? cloud-native technology.

By abstracting apps completely from
the OS and infrastructure layers,
In his excellent article, The Road to
containers provide unprecedented
Abstraction, Stephen O’Grady places
agility and flexibility to support a
containers on a continuum of virtualization
DevOps approach of continuous
that started in the 1960s with COBOL,
integration/delivery/deployment. In
which buffered program developers from
addition, container images launch
the bits and bytes of machine language.
much faster than VMs, making them
Another significant node on the
more suitable for today’s dynamic
virtualization continuum was Java
runtime environments, in which apps
Application Servers, which emerged at the
are expected to scale up and down on
turn of the millennium to provide an
demand.
abstraction layer between Java apps and
the underlying OS/hardware
infrastructure.

In parallel, we witnessed the rise of VMs,

with their ability to abstract and
standardize infrastructure deployments on
commodity servers. VMs are still widely
deployed in data centers around the
globe, and will most likely be with us for
many years to come. However, there are
numerous factors that have driven, and
will continue to drive, the growth of
container usage—whether alongside or
instead of VMs. These drivers include:

EDB® WHITE PAPER / PAGE 4 WWW.ENTERPRISEDB.COM

Containers save costs throughout the
application lifecycle, from dev/test to
production. Some quantified examples
of these savings (based on a Forrester
report) are:

Reduced hypervisor licensing fees that

can lower production environment costs
by as much as 50%.
Process optimizations that not only
streamline dev/test costs by as much as
70%, but also shorten time-to-market so
that revenues and other business
values are captured more quickly.
Thrifty runtime resource consumption
as compared to VM deployments, since
you don’t need to run as many copies
of the OS—up to 80% fewer servers.

As opposed to proprietary VM platforms,

containers have been primarily open
source. Enterprise development teams using
containers benefit from a vibrant open-
source community. They also avoid vendor
lock-in, with 83% of containers today
running on the royalty-free Docker Engine.
From a developer and operator standpoint,
container images run exactly the same in
development, staging, and production,
regardless of where they are deployed. This
consistency eliminates the “it works on my
machine” response that you often hear
when problems arise due to a discrepancy
between environments.

EDB® WHITE PAPER / PAGE 5 WWW.ENTERPRISEDB.COM

All of the benefits described above have OpenShift, GKE, Amazon EKS, and IBM
undoubtedly contributed to the Cloud Private, and is now being
enthusiastic adoption of containers. incorporated into frameworks that
However, we would argue that the single were not originally Kubernetes-based,
most important container value such as Docker Enterprise and Cloud
proposition is application portability, Foundry with PKS.
which is alluded to in the last bullet point.
With increasingly diverse and complex With orchestration, containers are
hybrid and multi-cloud environments
automatically reconfigured, scaled,
becoming the norm, it is a huge benefit to
upgraded, updated, and migrated
be able to operate the same software in
without disrupting applications and
the same way across multiple public and
services. The orchestration framework
private clouds, and even in on-premises
monitors the health of the runtime
virtualized environments. All the app
container deployment and ensures
deployment and infrastructure
high availability and continuity in
requirements are packaged in the
failover or disaster scenarios. It can
container image, which can be launched
also monitor load and automatically
anywhere, with full confidence that it will
scale services up and down in
perform as expected.
response to changes in demand. The
container orchestrator facilitates
It is also important to acknowledge that
networking among containers or to
containers could not have gone
mainstream without the emergence of external environments. It handles data

container orchestration frameworks that storage in general, and statefulness in

manage and deploy containers across particular, through the management of

clusters, on demand. The leading attached persistent or ephemeral

framework by far is Kubernetes, the open- volumes, which can be local or in the
source pioneer of container management cloud.
that originally came out of Google.
Kubernetes is also the underlying
foundation for other frameworks, such as

EDB® WHITE PAPER / PAGE 6 WWW.ENTERPRISEDB.COM

 The considerable disk space required to
Why store large amounts of data in a

Containerized containerized database makes it less

agile and less relocatable.

Databases Databases are inherently stateful and

durable, while containers are typically

Now? stateless and ephemeral. The

workarounds put into place to handle

A containerized database is an persistent data storage and longer-than-

encapsulation of its DBMS server software, usual container lifespans often detract

with access to a physical database file from the key container benefit of reduced

residing somewhere within the network. runtime resource usage.

Each DBMS is encased in its own container Databases typically have numerous

image. Containerizing a database, tuning parameters, many of which are

however, is not quite as straightforward as dynamic. Building a new immutable

containerizing an application. Some of the container image for every possible

database containerization challenges database configuration can quickly result

often cited are: in image sprawl. It should be noted,

however, that this issue is even more of a
Databases typically require high- challenge in VM deployments, since
throughput, low-latency networking. containers are considerably more
However, Docker containers do not lightweight than VMs.
natively provide the level of storage
and network resource isolation that is
necessary to achieve these
requirements.

An alternative to containerized databases is Database-as-a-Service

(DBaaS), an API-based cloud service model whereby the service
provider is responsible for the required database physical infrastructure
and server-side DBMS resources, including performance configurations.

EDB® WHITE PAPER / PAGE 7 WWW.ENTERPRISEDB.COM

The Bottom Line
Ultimately, each project team must make its own decision about whether a
containerized database or DBaaS is the best cloud database provisioning fit for the
overall application architecture and its development process. The primary drivers for a
DBaaS cloud deployment are typically productivity and simplified administration, while
for a containerized database, they are usually portability and automation. In any case,
when choosing the optimal pathway to a database in the cloud, the following
containerized database characteristics should be taken into account:

Smaller containerized databases are

well-suited to the microservices
architecture that is now often
favored over large monolithic
applications. In contrast to a DBaaS
database deployment, where the
database is a central resource for
multiple applications, a
containerized database becomes, in
essence, a component of the specific
application that it serves.
When the automated/scripted
deployment of containerized
databases is used in conjunction
with an orchestration framework, the
result works together with traditional
clustering to ensure high availability
for both stateful and stateless
systems.

EDB® WHITE PAPER / PAGE 8 WWW.ENTERPRISEDB.COM

 The inherent elasticity of containerized
databases supports more flexible, and
hence, less wasteful upfront database
capacity planning and provisioning. With
containers, you can approach the database
as an on-demand utility.
Similarly, because containerized databases
separate storage from compute, storage
performance and capacity can be scaled
independently of compute resources.
A software-defined containerized database
provides a crucial missing link in high-
velocity DevOps cycles, allowing
development and operations teams to
collaborate seamlessly.

One indication that containerized databases are trending is that Postgres, a well-
known open-source relational database, is currently the third most popular technology
being run on Docker.

Source: Datadog

EDB® WHITE PAPER / PAGE 9 WWW.ENTERPRISEDB.COM

How EDB Postgres
Accelerates
Containerization
Before we start laying out the architecture of a containerized Postgres deployment, it is
important to understand what a standard Postgres environment configured for high
availability looks like as illustrated below:

1. Monitoring and Administration: Each Postgres server is monitored by an agent that reports its tracked data to an
administration tool.
2. High Availability: EDB Postgres streams incremental changes from the master to any number of read-only standby
replicas. A high-availability agent watches for failures and automatically triggers failover protocols according to the
nature of the detected failure.
3. Disaster Recovery: EDB Postgres maintains a backup data store and automatically orchestrates enterprise-grade
disaster recovery procedures as required.
4. Scalability: EDB is one of the major contributors to the pgPool proxy, which helps applications scale by load
balancing read transactions to the replicas while directing write requests to the master. There is also a redundant
pgPool proxy to which the application is automatically connected by the high-availability agent in a failover situation.

EDB® WHITE PAPER / PAGE 10 WWW.ENTERPRISEDB.COM

In order to create a resilient Postgres architecture that can operate at scale, the
following four capabilities must be included:

Failover management for high availability

Database monitoring and administration management
Backup for disaster recovery
Query routing and load balancing for scalability

EDB Postgres is delivered as four containers, all of which are managed by Kubernetes:

1. Database Administration: Postgres Enterprise Manager® (PEM), packaged in this container, monitors the database,
collecting performance and status data that's displayed within dashboards and analyzed for alert conditions. These
alerts can be relayed to operators or to other enterprise-level management systems.
2. High Availability and Data Management: This container includes EDB Postgres Advanced Server along with EDB
Postgres Failover Manager (EFM). Users can have Kubernetes spin up several replicas of a master container so if the
master fails, one of the replicas can take over automatically.
3. Disaster Recovery: This container includes the EDB Postgres Backup and Recovery Tool (BART), which can back up
databases in multiple different containers. This enables BART to oversee multiple deployments.
4. Read Scalability: This container delivers query routing and connection pooling using pgPool. The benefit of this
container is scaling read activity where pgPool acts as a load balancer routing queries to replica databases. It sits in
front of the other containers and can be scaled independently of the database container.

EDB® WHITE PAPER / PAGE 11 WWW.ENTERPRISEDB.COM

Who Benefits From
Using Containers?
Anyone who wants to modernize their infrastructure.

EDB Postgres is platform agnostic and can failover. Replica pods are a way to
be deployed on any Kubernetes platform manage a deployment so that redundant
such as Red Hat OpenShift, Google database servers can be on separate
Kubernetes Engine, Pivotal Cloud Foundry, physical hardware.
Docker, IBM Cloud, or Amazon Web
Services (AWS). Management and monitoring is done
through PEM, with shared storage being
The following schematic shows a typical the preferable data storage practice in
configuration of EDB Postgres in a production for the database itself, as well
production environment, with a master and as for backup and recovery.
two replica pods for high availability and

EDB® WHITE PAPER / PAGE 12 WWW.ENTERPRISEDB.COM

EDB Postgres makes it possible for EDB customers to get immediate out-of-the-
box benefits from containerized Postgres databases. Typically, they start with
simple applications and move to more complex systems over time. EDB works
closely with its customers, sharing with them best practices for effectively
leveraging containers in next-generation, microservice-based applications.

Through its Architectural Roadmap and Solution Blueprint Service, EDB solution
architects help their customers design a customized container roadmap for
getting from where they are today (struggling with large and cumbersome
databases), to a new world of small nimble deployments that are fully aligned
with their strategic business objectives. Leveraging a reference library of proven
blueprints, full-stack tool sets, and API integration designs, EDB customers can
accelerate their digital transformation initiatives, model a robust open-source-
based data architecture, and deploy rapidly across complex environments.

EDB® WHITE PAPER / PAGE 13 WWW.ENTERPRISEDB.COM

Conclusion containerized databases have raised a
unique set of challenges in terms of high
data availability, backup and recovery,
In summary, containers and container
and other critical database performance
orchestration have matured to the point
and compliance requirements.
that they are now positioned at the very
core of cloud-native initiatives. Like VMs,
EDB Postgres allows enterprises to benefit
containers abstract the application layer
from containerized databases without
from the compute/storage/network
forfeiting critical database administration
infrastructure layer. Going one step
and monitoring requirements. Using native
further, however, containers also
Postgres Docker containers orchestrated
abstract the operating system layer.
with Kubernetes, EDB Postgres ensures
The result is a lightweight, resource-
high data availability, as well as seamless
optimized, self-contained application
failover, scalability and load balancing,
that performs consistently across a wide
automated monitoring and tuning, and
range of diverse platforms. Enterprises
robust backup and recovery.
around the globe are using containers
for their dev/test and production
workloads to rapidly develop and
deploy born-in-the-cloud products and
services that scale effortlessly.

In addition to containerized
applications, containerized databases
have emerged as part of the paradigm
shift from large monolithic applications Please visit the EDB website to
to applications composed of download and try out EDB containers
(click the Access Repository button).
microservices. Rather than a large
You can also learn more about the
centralized database that serves
supported options and platforms as
multiple applications, containerized well as view the documentation.
databases have become an on-demand www.enterprisedb.com/containers
utility that is an integral part of the
application itself. That being said,

EDB® WHITE PAPER / PAGE 14 WWW.ENTERPRISEDB.COM

EnterpriseDB | www.enterprisedb.com
EnterpriseDB, EDB and EDB Postgres are trademarks of EnterpriseDB Corporation.
Other names may be trademarks of their respective owners. Copyright© 2019. All rights reserved. 20190405