Design Principles for 


Multi-cloud Interoperability

emma.ms | Design Principles for Multi-cloud Interoperability 1

Contents
3 Introduction

Siloed vs. Interoperable Multi-cloud: Why Interoperability Matters?

4 Understanding Multi-cloud Interoperability

5 Designing the Multi-cloud for Interoperability

11 Enabling Interoperability in Multi-cloud with the emma Platform

emma.ms | Design Principles for Multi-cloud Interoperability 2

Introduction
Cloud computing platforms are not one-size-fits-all. That’s why an overwhelming
majority of organizations utilize more than one kind of cloud and cloud service
provider. Ideally, workloads running on any cloud should be able to communicate
and collaborate with each other. This ability of different systems and modules
to exchange data with each other across clouds is known as multi-cloud
interoperability. It’s one of the very basic attributes of cloud native systems.
Without it, much of the cloud’s scalability and agility would be lost. 

In our latest white paper, we discuss the key principles for ensuring
interoperability in diverse multi-cloud environments, which is not only
a prerequisite for optimal cloud utilization but also for fostering an ethical and
free market environment.

Siloed vs. Interoperable Multi-cloud:

Why Interoperability Matters?
Classically, a multi-cloud was any environment that utilized services from more
than one cloud provider. However, modern multi-cloud deployments have an
increasingly diverse portfolio of cloud services as well as cloud types, including
on-premise, private, and public clouds, all well-integrated to operate holistically
as a single entity. This shift from a disjointed multi-cloud to an integrated,
interoperable multi-cloud offers numerous benefits:

Seamless deployment, migration, and communication

1 4 Cross-cloud redundancy and disaster recovery
across all clouds

2 Localized cloud benefits and cost optimization 5 Unified view and control of security and management

3 Horizontal scalability

emma.ms | Design Principles for Multi-cloud Interoperability 3

Understanding Multi-cloud Interoperability
Before jumping into the design principles of an interoperable multi-cloud, it’s
essential to establish a clear distinction between interoperability and portability.
Often used interchangeably, they are, in fact, distinct concepts with nuanced but
significant differences. In the context of a multi-cloud:

Interoperability is the ability of two systems, Portability is the ability of applications and data to
1 2
components, or services running on separate clouds to migrate from one cloud to another without running
work together and communicate with each other into errors or losing functionality on the other
effectively without significant modifications. platform.

The concept of holistic interoperability has several layers to it.

Technical Interoperability Semantic Interoperability

The behavior of interconnected applications or components The data exchanged between multi-cloud applications or
must be consistent and predictable when integrated across services must be understood and utilized correctly.
clouds. Additionally, the protocols used for data transmission
must be reliable and secure across clouds.

Policy Interoperability

The rules, permissions, and access control policies for using

cloud resources must be consistent and compatible across
clouds to ensure adherence to organizational policies.

Collectively, these layers help achieve different aspects of interoperability in a

multi-cloud ecosystem.

emma.ms | Design Principles for Multi-cloud Interoperability 4

Designing the Multi-cloud for Interoperability
Interoperability is a fundamental attribute of the modern multi-cloud, fostering
a culture of fair competition, faster innovation, and social responsibility. Below
are the basic principles that ensure interoperability in data, applications, and
services. As such, they are all tightly knit and go hand-in-hand in enabling multi-
level interoperability.

API

Follow microservices- Use APIs as the default Use cloud agnostic Automate as much as Use CI/CD pipeline,
based design principles coding and minimize use possible to reduce agnostic to the cloud
for application design of platform-specific complexity provider, to increase
and consuption tools reliability and
repeatability

emma.ms | Design Principles for Multi-cloud Interoperability 5

Microservices-based Design

A microservices-based design decomposes complex applications into small, loosely

coupled, and modular components, each focusing on a single functionality. Each
microservice can be reused, deployed, and scaled individually, reducing
dependencies, complexity, and the need for rewriting the same code repeatedly.
However, there are several strategies and technologies that enable flexible,
adaptable, and independent microservices that are interoperable across any cloud.

Standardized Interfaces Containerization and Orchestration

Microservices must communicate with each other through Containers complement microservices and enhance their
standardized and well-defined APIs and protocols. APIs must portability and interoperability by encapsulating service code
use standard communication protocols, like HTTP/HTTPS for and its dependencies, such as the runtime, libraries, and system
RESTful APIs or gRPC for efficient remote procedure calls. This tools, into a lightweight, isolated, self-contained, and
ensures that microservices can communicate regardless of their executable package that can run consistently across any
technology stack, programming language, or underlying development, test, or production environment on any cloud. All
platform. Standardized APIs promote interoperability by major CSPs offer tailored container deployment and
exposing the functionality and data of each service in a orchestration services, like AWS Fargate, Amazon ECS (Elastic
consistent way, which allows microservices to understand and Container Service), and Azure Container Instances (ACI).
use each other’s functionality even when hosted on different However, in order to ensure multi-cloud interoperability,
clouds. containers must be deployed via a standard, vendor-neutral
container engine, such as Docker, and orchestrated through
platform-agnostic tools like Kubernetes.

Service Meshes Event-driven Architecture

Service meshes, like the open-source and platform-agnostic Event-driven architecture is a pattern that enables
Istio and Linkerd, provide advanced features and microservices to communicate asynchronously in response to
infrastructure abstractions for handling service-to-service events. This means that each microservice can operate
communication, service discovery, load balancing, security, and independently without waiting for the sequential execution of
observability between microservices. Service meshes can other tasks. Event brokers and platforms like Apache Kafka®
enhance the interoperability of microservices by assuming and RabbitMQ are responsible for distributing, storing, and
responsibility for network-level services and communication managing events, which decouples event-producing and event-
logic, thus boosting standardization and consistency in how consuming microservices. This decoupling enhances
microservices interact with one another. interoperability among distributed microservices, as they are
no longer directly dependent on other microservices to perform
their respective functions.

emma.ms | Design Principles for Multi-cloud Interoperability 6

Application Programming Interfaces (APIs)

Standardized communication is crucial for achieving interoperability in a multi-

cloud environment. APIs define how services and applications, with varying
underlying languages and technology stacks, communicate with each other
consistently. They serve as a layer of abstraction between applications and cloud-
specific services — instead of communicating directly with cloud-specific services,
applications can have their own standardized and generic APIs, which can then call
the cloud-specific API. This abstraction allows applications to communicate with
similar services from different clouds without needing significant modifications. 

However, there are certain prerequisites for ensuring that the APIs are broadly
compatible and interoperable.

Open Standards and Protocols API Documentation

APIs should conform to widely accepted standards and APIs must be accompanied by detailed documentation providing
protocols, for instance, adhering to RESTful principles, which information on endpoints, request/response formats, input
involve utilizing JSON or XML for data representation and using parameters, and error-handling mechanisms. Well-documented
HTTP/HTTPS to define actions (GET, POST, PUT, etc.) on APIs are easier for developers to understand and use across
resources. In addition, following practices like defining different platforms effectively. Tools like OpenAPI Specification
standardized endpoints and resource-oriented URIs ensures (OAS) or Swagger can help create machine-readable
consistency, understandability, and usability of the API across documentation for ensuring better standardization, enabling
clouds. automated integration, and facilitating consistent and error-
free interactions between systems and services across any
cloud platform.

API Gateways v.1 API Versioning

API gateways are servers or services that provide a centralized Most services with published APIs will need updates and
point for managing and controlling the flow of requests and modifications that will require changes to the API. It is
responses between microservices. They standardize inter- imperative to implement versioning in service APIs to ensure
service communication by translating between different consistency and backward compatibility. This ensures
protocols and data formats. They also handle request routing, interoperability between systems that may be using different
security, and authentication, thus abstracting the complexity versions of the API. It also allows seamless adaptability across
of the underlying microservices from API consumers. This multi-cloud platforms where different clouds may have varying
standardization and abstraction offers compatibility and update schedules and requirements. Tools like Swagger,
interoperability between services supporting different OpenAPI, and Postman allow version information in their API
technologies and platforms. documentation. Similarly, many API gateways and management
tools offer built-in support for versioning.

emma.ms | Design Principles for Multi-cloud Interoperability 7

Automation

Automation is the use of tools and scripts to streamline workflows and enable
actions, decisions, and responses without manual intervention. In a multi-cloud
environment, automation requires standardization of infrastructure,
configurations, deployments, and operations across different cloud environments.
Certain cloud-specific details must also be abstracted in order to maintain a
common language and approach to managing applications and resources in diverse
cloud environments.

Automation, through standardization, consistency, and abstraction, enables and

enhances interoperability in multi-cloud. Several aspects of cloud configuration
and management can benefit from automation as it boosts reliability, efficiency,
and interoperability.

Infrastructure as Code (IaC) Configuration Management

IaC abstracts the underlying infrastructure details and Configuration management tools like Ansible, Chef, and Puppet
specifications, defining it in a version-controlled and machine- abstract the underlying cloud-specific details, enabling
readable format that is consistent and reproducible across organizations to define cloud-agnostic configuration playbooks
different cloud environments. Tools like Terraform and Puppet for configuring servers, applications, and infrastructure
enable organizations to define and provision cloud resources resources. These playbooks can then be run across any cloud to
consistently across multiple clouds. automate configurations and yield consistent results.

Container Orchestration Continuous Integration and Continuous Deployment (CI/CD)

Container orchestration tools like Kubernetes automate the CI/CD pipelines are a key part of the DevOps strategy. They
deployment and scaling of containerized applications. They automate build, testing, and deployment of applications and
abstract cloud-specific details and provide a consistent way to services for faster and more reliable software delivery.
deploy and manage containers across different clouds, making it Carefully implemented CI/CD pipelines support interoperability
easier to achieve interoperability in multi-cloud. by promoting consistency and enabling automated, early
detection of compatibility issues that can hamper
interoperability.

emma.ms | Design Principles for Multi-cloud Interoperability 8

Cloud Agnosticism

Cloud agnosticism ensures that an application or system remains independent of a

specific cloud provider. Cloud agnosticism, as a design and operational approach,
can significantly boost interoperability through open standards, abstractions, and
consistency in deployment. As such, it requires avoiding vendor-specific tools and
services and embracing open standards and cloud agnostic tools and technologies
at multiple layers of an application and system design, development, and
deployment.

Code and Development Infrastructure

At its core, interoperability demands that applications be At the infrastructure level, cloud agnosticism involves
written using languages and frameworks that are cloud designing applications to run and communicate consistently on
agnostic. This means developers should avoid cloud-specific any cloud's virtual machines or containers. Vendor-neutral IaC
APIs, SDKs, and services and opt for open-source libraries and tools like Terraform and Ansible can be used to define cloud
tools that work consistently across cloud platforms. Using agnostic infrastructure configurations, thus reducing
RESTful APIs, gPRC, and other widely accepted protocols for dependencies on cloud-specific features and enabling
application interactions promotes interoperability. applications to communicate and collaborate seamlessly in a
distributed, multi-cloud setting.

Data Storage Cloud Services

To ensure interoperability, data storage solutions should not be In addition to databases and IaC tools and services,
tightly tied to any particular cloud provider. For instance, organizations should opt for open-source and/or vendor-
instead of using AWS DynamoDB as a No-SQL database, neutral versions of any managed offerings they need, for
organizations should opt for an open-source distribution like instance, choosing Kubernetes instead of AWS EKS, Azure
MongoDB or Redis, which are widely supported and can interact Kubernetes Service (AKS), or Google Kubernetes Engine (GKE).
with applications across different platforms without code This ensures that containerized applications across different
refactoring. clouds remain interoperable. In cases that necessitate the use
of a proprietary service, the service should be consumed via API
abstractions.

Management Tools

Major cloud providers are now offering multi-cloud platforms,

like Google Anthos, Azure ARC, and IBM Cloud Satellite, but they
often restrict organizations to the clouds and environments
they support. For instance, Google Anthos offers limited
functionality when it comes to managing clusters in Microsoft
AKS and Amazon EKS. To fully use Anthos, administrators must
migrate their workloads to Google's Kubernetes engine. Such
limitations make multi-cloud management platforms from
hyperscalers less interoperable as opposed to cloud agnostic
multi-cloud management platforms, like the emma platform.
The emma platform supports various containerization
technologies and extends container management and
orchestration capabilities to all environments — any cloud,
edge, or on-premise.

emma.ms | Design Principles for Multi-cloud Interoperability 9

Vendor-Neutral CI/CD Pipeline

In keeping with the principle of vendor agnosticism, it’s important to embrace

open-source and vendor-neutral tools throughout the CI/CD pipeline. The idea is
that organizations must plan for interoperability right from the early stages of
development. The CI/CD pipeline itself has several stages, each serving a specific
purpose in the software development and delivery process and utilizing its own
set of tools and technologies. A vendor-neutral CI/CD pipeline entails that none
of the tools and technologies utilized across the pipeline are tied to any specific
cloud or technology vendor. This ensures that the CI/CD process can work with
diverse technologies, cloud platforms, and infrastructure configurations.

Below are the key stages of a CI/CD pipeline and how each can be made vendor-
independent to ensure consistency and interoperability when different phases of
the pipeline are executed on different clouds or when different components of the
same application are deployed across different clouds:

Source Code Repository Continuous Integration

The source code must be stored on a cloud-agnostic version Tools like Jenkins, GitLab, or CircleCI are needed to build and
control system like Git that integrates seamlessly with any test the application continuously. They connect seamlessly with
cloud platform or tools needed for the subsequent stages. Git and work across multiple cloud providers.

Deployment

Packaging applications into containers using Docker, an open-

source containerization platform, maintains portability. It goes
hand-in-hand with Kubernetes, which is vendor-neutral itself
and automates the deployment and coordination of Docker
containers in complex, large-scale containerized applications
distributed across multi-cloud clusters.

emma.ms | Design Principles for Multi-cloud Interoperability 10

Enabling Interoperability in Multi-cloud
with the emma Platform
Enabling an interoperable multi-cloud requires careful planning from the get-go.
However, ensuring interoperability by design may not be feasible for many
enterprises in different phases of their multi-cloud journey. emma is a cloud-
agnostic, no-code multi-cloud management platform that takes the complexity
out of multi-cloud interoperability. It enables different environments, legacy on-
premise, private, public, and multi-cloud, to integrate and work together
seamlessly through abstractions, single-pane-of-glass visibility, and intercloud
connectivity. These attributes go beyond technical interoperability to foster
semantic and policy interoperability as well, creating a cohesive and interoperable
multi-cloud. 

Here’s how the emma platform facilitates and enables organizations to embrace
the seemingly complex principles of multi-cloud interoperability:

Microservices Architecture Cloud-specific APIs

The emma platform facilitates seamless communication between The emma platform ensures consistency in deployment
microservices across clouds by abstracting the complexities of mechanisms when dealing with cloud-specific APIs and
different cloud interfaces. Its no-code approach allows admins configurations. It abstracts the complexities and variations of
to deploy virtual machines and containers across multiple individual cloud APIs and provides a unified and standardized
clouds in just a few clicks. 
 deployment framework. The platform incorporates a powerful
The platform’s load-balancing capabilities that span the entire API gateway that serves as a centralized hub, enabling smooth
multi-cloud ecosystem, allowing efficient traffic distribution communication across diverse systems and cloud providers.
and optimal resource utilization across all clouds.

Automation Cloud Agnosticism

The emma platform's Kubernetes capabilities allow automated The emma platform abstracts cloud-specific interfaces to
container orchestration across multiple clouds. In addition, enable users to deploy workloads on any cloud platform in a
users can create policies and rules to automate cloud tasks and cloud agnostic manner. The platform itself is cloud agnostic,
workflows, such as infrastructure resource provisioning, cross- which means it integrates seamlessly with all public, private,
cloud application scalability, compliance checks, and backups and on-premise cloud deployments. Organizations can manage
and disaster recovery. and integrate their applications and infrastructure regardless
of where they are hosted, all through a single platform.

Vendor Agnostic CI/CD Pipelines

The emma platform supports automated testing, deployment,

and integration mechanisms through integrations with various
CI/CD tools. Organizations can integrate their CI/CD tools,
such as Jenkins and ArgoCD, to to ensure consistent deployment
pipelines, automation workflows, and version control in a multi-
cloud environment.

emma.ms | Design Principles for Multi-cloud Interoperability 11