Putting cloud repatriation into perspective

Moving cloud workloads back on-premise may seem like backpedalling on an organisation’s cloud journey but those that do so are usually trying to rein in escalating cloud costs.

According to Matt Young, senior vice-president and head of Nutanix Asia-Pacific and Japan, enterprises like the flexibility of public cloud, but may find it too expensive over time, spurring them to look for the same capabilities through private cloud infrastructure.

Young said cloud repatriation is happening based on his interactions with customers, but the extent to which it is taking place is unclear. According to a Gartner study in 2019, public cloud repatriation remains the exception rather than the rule as cloud consumption continues to grow.

Jim Freeman, IBM’s chief technology officer for global technology services in Asia-Pacific, said enterprises that have repatriated cloud workloads tend to have a less robust workload placement strategy and lack cloud management skills.

He said enterprises should have a cloud workload strategy that identifies the functional requirements for an application. Equally important, he added, are non-functional requirements such as data gravity, or the ability of data to attract applications, services and other data.

“Cloud repatriation is often the result of two things: a lack of understanding that you should be turning things off when you’re not using them, and the failure to realise the data consumption rate of an application,” said Freeman.

Other common drivers for repatriating cloud workloads include additional security and compliance demands imposed by regulators, cloud outages that impact service-level agreements and skills gaps that could have security and performance implications.

In reality, cloud adoption remains a continuum, with enterprises straddling between public and private cloud environments hosted in on-premise datacentres. Singapore’s DBS Bank, for example, operates a private cloud while making use of Amazon Web Services (AWS) to host its public web assets.

“AWS claimed that only 5% of all the compute in the world is on the public cloud,” Young said. “You can look at that in one of two ways: there’s a huge opportunity for public cloud but at the same time you can also say that 80% of workloads are better suited from a performance and cost perspective on the private side, and so it’s very much a hybrid environment,” Young said.

In Singapore, local IT decision makers reported plans to boost their use of hybrid cloud by 54 percentage points, according to Nutanix’s latest Enterprise cloud index. Those who use private cloud only (32%) is also higher than the Asia-Pacific average of 23%.

Whichever cloud environment an organisation chooses, whether it’s public, private or hybrid, the traditional three-tier architecture no longer suffices. Besides adopting a cloud-like architecture, organisations will need to design workloads so that they be moved between public and private clouds.

This portability of workloads has been enabled through containerisation and Kubernetes, which can be found in nearly every enterprise, in private cloud datacentres and on public clouds, wherever microservices-based applications are deployed.

Cloud, infrastructure and application platform suppliers, including the likes of AWS, Nutanix, VMware and Red Hat, are all jostling for a piece of the market that is slated to grow to nearly $5bn in 2023, according to some estimates.

At the end of the day, enterprises need to get smart about their use of cloud, Young said. “You can’t jump headfirst; you have to be smart about it and where the data is going to live is a big part of it.

“For example, you want to put data into a public cloud, and you want to keep it local but there are also issues with latency. Everybody’s on a journey and what people are looking for is a cloud-like experience in a hybrid model.”