Living in a multicloud world

Inexorably, the world has become a multicloud one, and it’s now widely accepted that any organisation actively pursuing a cloud journey – and that’s the vast majority – will have a strategy for deploying their workloads into multiple clouds. But there still could be some lingering confusion around this, so to clear this up from the outset, you’re a multicloud organisation if you’re using more than one cloud for similar IT solutions.

There are a number of contributing factors to multicloud adoption. Eugene de Souza, regional cloud and service providers lead, Sub-Saharan Africa, Red Hat, says multicloud in, for example, financial services, will probably be driven by regulatory concerns, as well as business continuity, risk mitigation and the need for high availability.

“Do all industries need multicloud? I think it depends on the criticality of the application or the service that’s being consumed. It’s also important to have some form of backup, or have an alternative [cloud] for any customer-facing applications that are revenue-generating or mission-critical. Diversification is one driver, but the hypercompetitive nature between the cloud providers is also driving companies to adopt multicloud strategies,” says De Souza.

So much for the drivers, but it’s not a good idea for a company to jump into multicloud straight away and the complexities may well prove its undoing. Managing multicloud is more complex than just a single instance, and will also mean you’ll need a larger talent pool. De Souza says it’s very rare to find a person who knows their way around both AWS and Azure, and is able to keep up to date with the certifications and new features. It’s more likely that there’ll be two teams of people, and this will come at a cost.

Jon Tullett, research manager, IT services, IDC Sub-Saharan Africa, says if a customer is embarking on a cloud journey, it will almost certainly have to turn to a third-party professional services organisation, and that the local market is starting to coalesce around larger channel partners.

Multicloud will also probably cost more. De Souza says a company only using a single provider will be incentivised by discounts based on its usage. If that spend is split across two public clouds, it’ll have less leverage and bargaining power because its budget is split in two.

The public cloud provides very high availability, and firms in South Africa are now spoiled for choice with AWS and Azure regions, not to mention Google’s cloud, Alibaba, Huawei and Oracle. And while these all have typically high availability, they do go down every now and again.

What about app portability between clouds?

De Souza says this may require some refactoring, or tweaking. Should a customer develop an application on a native service in one of the cloud providers, and it doesn’t allow for portability, that application will then be locked into that native service.

At the moment, there’s a significant difference between the advanced AI and analytics you can get on one platform versus another.

Jon Tullett, IDC

As for software development, there are a number of enterprise Kubernetes deployment options available, such as SUSE’s Rancher, VMware’s Tanzu, or Red Hat’s OpenShift.

De Souza says Red Hat has teamed up with both AWS and Azure to run its app development platform.

Tullett says he’s seeing a tremendous amount of local interest in private cloud, both on-premise and hosted. He adds that one probably won’t see large application fleets spanning multiple clouds unless it’s for the sake of redundancy.

“Generally, you can find most of what you need within a specific platform, so you’ll be leveraging platform A for one application stack and all its related pieces, and a separate platform for platform B because it was a better fit, and then you’ll use all additional components within that platform to support it.

“You typically won’t see applications breaking out and leveraging components in separate clouds unless there’s a very good reason for it, and there are sometimes good reasons,” he says, such as in the case of an ERP, which may have third-party modules resident in different clouds.

“At this point, multicloud is really more a conversation around how to apply policy across multiple clouds simultaneously for the sake of efficiency and uniformity. And that’s really important in areas such as cost-optimisation because there may be specific areas where one cloud has a cost performance advantage over another. And for security, where you’re trying to deploy a security policy, you need to make sure that it’s coordinated across every environment, whether that’s cloud or not.”

More of the hybrid and multicloud conversations right now tend to be around managing multiple separate cloud environments, but that will change in the future, Tullett says.

“Part of the evolution towards a more complex PaaS is that ability to actively manage multiple clouds and start to leverage within the context of single application environments.

“One reason for this is that the application is getting more complicated, and, secondly, there’s a whole raft of advanced solutions coming down the line where cherry picking is very desirable,” he says, mentioning AI as one example.

“At the moment, there’s a significant difference between the advanced AI and analytics you can get on one platform versus another. If you want to take advantage of that in an application that’s on a different cloud, you will break out, and you ‘ll be looking to integrate from a third-party into that.”