Amid the need for speed, agility, and scale capabilities, a multi-cloud strategy is high on the enterprise agenda. However, it’s also notoriously difficult to get right. Despite having so much promise, many enterprises can tell discouraging stories about their multi-cloud challenges, including spiraling costs, lack of control, poor visibility, overly complex and time-consuming setups, and resource intensity.
So, how can multi-cloud be made ‘better’?
How multi-cloud is misunderstood
Multi-cloud obstacles are often born out of inadequate approaches, both on the vendor side and enterprise side.
Many vendors claim to provide multi-cloud networking, but they do so in a very narrow sense. Their bread-and-butter is to provide connections to the edge of each cloud before handing off the networking to the cloud provider's native constructs. A useful analogy could be to imagine building a highway and all the exit points but forgetting about the local roads to get you to the actual destination. That is the experience you get with many vendors.
Simultaneously, many enterprises believe that having two or more cloud providers is what makes them multi-cloud. While true in a literal sense, having 2+ cloud providers does not make what are largely disconnected, and incompatible clouds appear as one. Without a proper approach, having multiple clouds creates challenges around the complexity and different skills needed to build and manage islands of disparate networks for each cloud service provider (CSP).
In another helpful analogy, imagine owning two adjacent apartments. Only when you tear down some walls and join the apartments will you get one large functioning apartment. Otherwise, you still have two distinct apartments that you maintain separately. That’s the difference between using multiple clouds and being truly multi-cloud connected.
Vendors and enterprises could be forgiven for not understanding the full scope of multi-cloud networking; after all, the concept is relatively new. However, the result in both scenarios is a failure to provide consistency and repeatability across cloud networking, network security, operational visibility, troubleshooting, and automation. Without that consistency, enterprises will fail to maximize the agility that should come with leveraging CSPs.
Unlocking the benefits
At its most basic, multi-cloud networking could be defined according to these five best practice principles.
- It has to be an on-demand service, distinct from an on-prem capability built out of the enterprise data center.
- It should provide end-to-end network connectivity to facilitate communication to and between applications and workloads across clouds, cloud services, on-premises data centers, and edge networks (a mid-mile or last-mile-only service will not suffice).
- It should offer fully integrated network and network security services (such as a firewall and protection from bad actors) with an easy way to provision new third-party services.
- It should provide end-to-end multi-cloud governance and multi-cloud visibility.
- It should not be restricted to a single cloud provider. Instead, it should enable secure connectivity to multiple clouds and edge locations.
To unleash the full potential of multi-cloud networking, you should think about how you will achieve and maintain consistent and optimal networking and network security, consistent visibility, consistent troubleshooting, and consistent Infrastructure as Code (IaC) for any cloud. In other words, you want to build once and repeat many times.
While this doesn’t happen naturally, the use of a cloud-native overlay to provide that consistency and simplicity becomes even more important to avoid the complexity of handling different clouds and cloud constructs.
A single front-end solution
A single front-end — one where you can spin up workloads or migrate services to any cloud in a rinse-and-repeat manner – will be a great source of relief to cloud IT teams. That way, you can create and manage your own network to cloud and within the cloud using just one console for management irrespective of how many clouds are needed.
Such a solution can control and orchestrate the various underlying constructs so that you can launch and manage your VPCs and VNETs in a consistent way and overlay these with a superset of capabilities that extend across all clouds. Needless to say, you will save your teams a lot of time, stress, and headache if you take a consistency-driven approach to your cloud network.
The important lesson is true multi-cloud networking describes a world in which you no longer have to perform AWS activities 'the AWS way,' or GCP activities 'the GCP way' and so on. Instead, you can do it your way by connecting your clouds using a multi-cloud network architecture.
John Gonsalves is Principal Field Evangelist at Aviatrix.