Your submission was sent successfully! Close

Thank you for signing up for our newsletter!
In these regular emails you will find the latest updates from Canonical and upcoming events where you can meet our team.Close

Thank you for contacting our team. We will be in touch shortly.Close

  1. Blog
  2. Article

Tytus Kurek
on 27 May 2020


Hybrid cloud and multi-cloud are two exclusive terms that are often confused. While the hybrid cloud represents a model for extending private cloud infrastructure with one of the existing public clouds, a multi-cloud refers to an environment where multiple clouds are used at the same time, regardless of their type. Thus, while the hybrid cloud represents a very specific use case, multi-cloud is a more generic term and usually better reflects reality.

Although both architectures are relatively simple to implement from the infrastructure point of view, the more important question is about workloads orchestration in such environments. In the following blog, I describe the differences between hybrid clouds and the multi-cloud and discuss the advantages of orchestrating workloads in a multi-cloud environment with Juju.

Understanding the difference between the hybrid cloud and the multi-cloud

Let’s assume that you manage a transport company. The company owns ten cars which is sufficient in most cases. However, there are days when you really need more than ten. So how do you handle your customers during those traffic-heavy periods? Do you buy additional cars? No, you rent them instead. You rely on an external supplier who can lend you cars on-demand. As a result, you can continue to deliver your services. This is almost exactly how the hybrid cloud model works.

However, the reality is slightly different. First of all, companies never rely on a single supplier. You need at least two suppliers to ensure business continuity in case the first one cannot provide their services. Moreover, not all cars are the same. What if you need a really big one which none of your existing suppliers can provide? You would probably rent it from another supplier, even if this only happens once a year, wouldn’t you? And finally, it is quite possible that even though you are in need of buses, due to lack of experience with them you are not really willing to own and maintain them, even if in the long term that would have been a more cost-efficient option.

The second case represents the multi-cloud model and is closer to what we are observing in modern organisations. While the hybrid cloud concept was developed as a solution to offload a private cloud during computationally-intensive periods, an orchestration of workloads in multi-cloud environments is what most organisations are really struggling with nowadays. This is because not hybrid clouds, but the multi-cloud is a part of their daily reality. 

Hybrid cloud and multi-cloud: architectural differences

A hybrid cloud is an IT infrastructure that consists of a private cloud and one of the available public clouds. Both are connected with a persistent virtual private network (VPN). Both use a single identity management (IdM) system and unified logging, monitoring, and alerting (LMA) stack. Even their internal networks are integrated, so in fact, the public cloud becomes an extension of the private cloud. As a result, both behave as a single environment which is fully transparent from the workloads’ point of view.

The goal behind such an architecture is to use the public cloud only if the private cloud can no longer handle workloads. As the private cloud is always a cheaper option, an orchestration platform always launches workloads in the private cloud first. However, once the resources of the private cloud become exhausted, an orchestration platform moves some of the workloads to the public cloud and starts using it by default when launching new workloads. Once the peak period is over, the workloads are moved back to the private cloud, which becomes the default platform once again.

In turn, multi-cloud simply refers to using multiple clouds at the same time, regardless of their type. There is no dedicated infrastructure that facilitates it. There is no dedicated link, single IdM system, unified LMA stack or an integrated network. Just instead of a single cloud, an organisation uses at least two clouds at the same time.

The goal behind the multi-cloud approach is to reduce the risk of relying on a single cloud service provider. Workloads can be distributed across multiple clouds which improves independence and helps to avoid ‘vendor lock-in’. Furthermore, as the multi-cloud is usually a geographically-distributed environment, this helps to improve high availability of applications and their resiliency against failures. Finally, the multi-cloud approach combines the best advantages of various cloud platforms. For example, running databases on virtual machines (VMs) while hosting frontend applications inside of containers. Thus, workload orchestration remains the most prominent challenge in this case. 

Orchestrating workloads in a multi-cloud environment

When running workloads in the multi-cloud environment, having a tool that can orchestrate them becomes essential. This tool has to be able to provision cloud resources (VMs, containers, block storage, etc.), deploy applications on top of them and configure them so that they can communicate with each other. For example, fronted applications have to be aware of the IP address of the database, as they are consuming the data stored there. Moreover, as the resources are distributed across various cloud types, the entire platform has to be substrate-agnostic. Thus, the entire process needs to be fully transparent from the end user’s perspective.

One of the tools providing this kind of functionality is Juju. It supports leading public cloud providers as well as most of those used for private cloud implementations: VMware vSphere, OpenStack, Kubernetes, etc. Juju allows modelling and deployment of distributed applications, providing multi-cloud software-as-a-service (SaaS) experience. As a result, users can focus on shaping their applications, while the entire complexity behind the multi-cloud setup becomes fully abstracted.

Summary

Although looking similar at first sight, hybrid cloud and multi-cloud are two different concepts. While hybrid clouds focus on offloading private clouds, the multi-cloud approach attempts to address the challenges associated with using multiple clouds at the same time. The biggest challenge with multi-cloud is not the infrastructure setup. It’s workload orchestration. Juju solves this problem by providing a multi-cloud SaaS experience.

Learn more

Canonical provides all necessary components and services for building a modern private cloud infrastructure. Those include OpenStack and Kubernetes, as well as Juju – a software for workloads orchestration in multi-cloud environments.

Get in touch with us or watch our webinar – “Open source infrastructure: from bare metal to microservices” to learn more.

Related posts


Tytus Kurek
3 April 2024

OpenStack with Sunbeam as an on-prem extension of the OpenStack public cloud

Cloud and server Article

One of the biggest challenges that cloud service providers (CSPs) face these days is to deliver an extension of the public cloud they host to a small-scale piece of infrastructure that runs on customers’ premises. While the world’s tech giants, such as Amazon or Azure, have developed their own solutions for this purpose, many smaller, ...


Andreea Munteanu
30 January 2024

AI on-prem: what should you know?

Cloud and server Article

Organisations are reshaping their digital strategies, and AI is at the heart of these changes, with many projects now ready to run in production. Enterprises often start these AI projects on the public cloud because of the ability to minimise the hardware burden. However, as initiatives scale, organisations often look to migrate the workl ...


Tytus Kurek
20 May 2024

Cloudify your data centre – A guide to VMware infrastructure transformation

Cloud and server Cloud

You know what’s going on. You’ve been monitoring the situation around VMware for at least a year now. There is no need to convince you that whatever comes next, you have to prepare for a big change. You and your team are already well prepared. You have a budget, timeline and necessary resources. However, one ...