Sergey Nivens -

Making multicloud work: Is it really worth it?

Enterprise interest in multicloud deployments is on the rise, but with so many moving parts to take care of, will the benefits end up being too hard-fought to make it viable for most enterprises?

The promise of multicloud suggests enterprises should be able to run their applications and workloads in whichever cloud environment makes the most sense from a cost, performance or functionality perspective at that moment in time.

If the cost economics of running a particular application in a certain cloud suddenly becomes unfavourable, enterprises should have the freedom to shunt them off to someone else’s with minimal fuss and effort, as needs dictate.

At least in theory, that is how multicloud is supposed to work, but the reality of the situation can be very different in practice, as enterprises grapple with how best to make technologies created by competing suppliers play nicely together.

“The more platforms you have, the more complexity you get – creating far greater challenges at the management level,” says Alex Dalglish, UK services director at IT services consultancy SoftwareOne.

“From cloud-sprawl and unplanned expenditure, to poor performance and availability or increased security risk – the diversity in your cloud platforms is directly proportional to the amount of supplier management burden on the IT team.”

Stephan Fabel, director of products at open source software company Canonical, backs this view, saying that the complexities involved with managing a multicloud deployment can be off-putting to some enterprises.

“The biggest cause for hesitation around multicloud adoption is that deploying cloud platforms and services from multiple vendors is complicated, especially when it comes to pulling them together in a way that doesn’t hinder productivity or innovation,” says Fabel.

What also does not help is the fact that multicloud is not a deployment model that every player in the public cloud has been willing to acknowledge or even pay lip-service to, despite rising end-user interest. 

“The more platforms you have, the more complexity you get – creating far greater challenges at the management level”
Alex Dalglish, SoftwareOne

The Google Cloud team are a notable exception here, having talked openly for several years about how common it is for its off-premise technologies to be adopted by enterprises that already operate a large Amazon Web Services (AWS) footprint. Sometimes this is done to gain performance, resiliency or to reduce operating costs.

For AWS, the multicloud message does not really square with its push to get more and more of its customers to declare themselves “all-in” on its platform. Microsoft seems to be following a similar line, while also thinking that multicloud might be a relatively short-lived strategy for some organisations.

“We believe that, over time, organisations will most likely embrace and choose a primary cloud provider and stick with it to leverage the services and ecosystem fully,” wrote John M Clark, a cross-domain solutions architect at Microsoft, in a blog post from October 2018. “Especially as applications are modernised or re-factored to take full advantage of that cloud’s platform-as-a-service (PaaS), function-as-a-service (FaaS) and software-as-a-service (SaaS).

“Some will settle on a couple of clouds as they are big enough to invest the time and people to go deep on both, and that makes sense if the clouds offer differentiated value beneficial to the workload, but only if you go in deeply with both and take advantage of the PaaS services that each offers.”

With this in mind, it is perhaps no surprise that Google is the first of the public cloud big three to have brought to market a multicloud management platform to help enterprises shift workloads between its cloud and Amazon and Microsoft-hosted environments.

Anthos, as the platform is known, is built using open source technologies, the most notable of which is the container orchestration engine Kubernetes, which forms part of the wider Google Kubernetes Engine (GKE).

The latter is a managed environment that allows enterprises to run containerised applications in on-premise and cloud environments without needing to create their own Kubernetes clusters first.

In the context of Anthos, it enables enterprises to containerise their applications and virtual machines for reasons of portability, and move them between the Amazon, Google and Microsoft clouds.

According to Leighton James, CTO of public sector-focused cloud provider UKCloud, investigating container technologies should really be the first port of call for any organisation looking to achieve workload portability across multiple clouds.

“For example, OpenShift from Red Hat is a Docker and Kubernetes-based platform that runs across Amazon, Google, Microsoft and other clouds,” he says.

“Many organisations will have deployed natively onto a global cloud platform, in which case there are two main options. VMware and Microsoft both have tools that you can use to move workloads from their public cloud to their private stack, and there are third-party tools such as Zerto which provide data replication and migration tools.”

Making multicloud work

Despite what some members of the public cloud community may choose to believe, there are solid reasons why users might want to supplement their AWS deployment, for example, with technologies from one of its competitors, says Dave Locke, chief technology adviser for Europe, Middle East and Africa (EMEA) at IT consultancy World Wide Technology (WWT).

“Each of the cloud providers excels in different areas,” he says. “AWS is great for the consumer solution, Azure – backed by Microsoft’s lengthy heritage – works well for enterprise, and Google Cloud Platform has superb big data and analytic tools.

“Firms understand these different strengths and line them up with what they want to achieve. This, in turn, provides an outline of what their ideal [multicloud] deployment should look like.”

Straight out of the gate, financial services company HSBC and Keybank came out as early adopters of Anthos, which has been designed by Google to provide a consistent user experience across all three public clouds.

This is important because a common stumbling block in multicloud deployments is sourcing staff to manage these setups who have enough in-depth knowledge and experience in using the tools and technologies of multiple providers, rather than just one, says WWT’s Locke.

“Most businesses are very skilled when it comes to working with their on-site servers or privately hosted solutions,” he says. “However, when it comes to utilising the public cloud, there is a significant knowledge gap.

“Simply put, most companies don’t have the skills or the resources to successfully create a deployment involving multiple public cloud providers. The ever-evolving nature of the public cloud can provide internal teams with a similarly fluid challenge.”

According to UKCloud’s James, when it comes to sourcing the right talent to manage a multicloud deployment, the best way to start is to develop the skills and draw on the experience that enterprises already have at their disposal in-house.

There are, after all, well-documented skills shortages pertaining to cloud in the UK, and hitting the recruitment trail in pursuit of people who can deliver on multicloud might be something of a fool’s errand.

“The key to handling multiple workloads is to harness the teams’ existing skills, experience and knowledge that’s been built when working with existing technologies like VMware, Oracle and Cisco,” says James.

“Steadily modernising an enterprise’s datacentre by gradually adopting new approaches like DevOps and agile is recommended, rather than integrating all of these in one fell swoop. Often, that’s where problems occur.”

Multicloud management problem areas

Aside from having to manage three disparate cloud environments, organisations also need to work out how best to juggle invoices from several providers, as well as keeping tabs on any changes that might occur to their security policies and ensure these environments are optimised for each workload running in them at all times.

On top of that is the added pressure of keeping on top of all the new features and functionality providers are liable to add to their platforms, as none of the big three are slouches when it comes to rolling out changes and new services.

And for an IT department looking to make a go of multicloud, keeping up to speed with whose technology does what in which cloud environment is tricky, says Lee James, CTO for EMEA at managed cloud provider Rackspace.

The company has undergone some huge changes in recent years as the public cloud market has matured. These have seen the firm pivot away from being a competitor to Amazon, Microsoft and Google in the infrastructure-as-a-service (IaaS) market, to becoming a collaborator through the roll-out of its managed services offerings.

According to James, about 10% of Rackspace customers are operating IT environments containing one private cloud plus two or more public clouds. One of the biggest challenges they face when trying to make these setups work is keeping up with all the innovation going on in the public cloud side of these deployments.

“The most common challenge to navigate when running a multicloud environment is around management and keeping your eyes on the prize across different providers with thousands of regular updates,” he says.

“AWS released 400-plus product updates last quarter. Extrapolate that across your multicloud environment and that’s a lot to manage. This leads to what we call the ‘magpie effect’, where the sheer volume of product releases has businesses constantly chasing the latest update and ultimately losing sight of their core requirements.”

At the same time, adopting a multicloud approach can bring the added benefit of making it much easier for organisations to gain access to the latest and greatest technologies the public cloud providers have to offer, says Canonical’s Fabel.

“Different vendors will innovate in different areas, which means businesses taking a multicloud approach can exploit developments as soon as they become available, rather than having to wait for one vendor to catch up,” he says.

Read more about multicloud

According to Locke, as well as keeping up to speed with the new features and functionality being introduced across the various cloud platforms they use, enterprises also need to have systems in place to ensure they can keep tabs on any issues and outages, too.

“Oversight of the multiple platforms can be an ongoing challenge,” he says. “It’s not just about fixing issues and outages, because companies need to be fault-finding and stopping issues before they become business-critical.

“With the right management, this is eminently achievable, but as multiple vendors all push their performance capabilities, it can be hard for companies to truly understand how the deployment will work until it is up and running.”

One workaround for this type of problem, says SoftwareOne’s Dalglish, is to source a management layer that offers a high degree of visibility into the inner workings of each of the platforms that make up the enterprise’s multicloud deployment.

“Enterprises need an overarching management layer that will give CIOs real-time insight into spend and performance,” he says. “IT departments must also make sure controlled management processes are enforced, so they control procurement of any new cloud service that a line of business is looking for. Taking these steps will ease the strain on IT teams and ensure that substantial cloud investments deliver ROI [return on investment].”

All things considered, for some organisations, multicloud might end up being more hassle than it is worth, so enterprises will have to weigh up for themselves whether the extra administrative burden involved is balanced out by any cost, performance or resiliency improvements they might gain.

Read more on Infrastructure-as-a-Service (IaaS)

Data Center
Data Management