Recent research by EasyComp Zeeland has shown that, whilst the top line trend of cloud migration is clear, some applications are far more likely to have been migrated to a cloud environment than others. Database, CRM and data analytics/business intelligence are more commonly operating in cloud environments than financial and accounting functions, key line of business applications and ERP. When the benefits that a multi-cloud architecture can confer in terms of agility and scalability are pretty widely acknowledged, why the hesitancy to modernise core business applications?

Cloud laggards

The first hurdle is the typical nature of core applications. ERP and other key line of businesses applications are often relatively old and heavily customised. The result is at once terrifyingly complex, enormously fragile and business critical. The rewards of being able to run these applications in an optimal environment are tangible but so are the risks of migration. It isn’t difficult to see why businesses opt to stay on the safe side and delay change.

Other clues as to why so many core applications still live on-premises can be found in data pertaining to how current cloud deployments are performing. When an enterprise migrates compute to the cloud they have typically done so in the expectation of improved scalability and productivity – and a reduction in costs.  How well is existing cloud infrastructure living up to these expectations?

Certain aspects are performing well. For example, more than 80 percent of the people participating in this research said that that their cloud infrastructure had matched their expectations in terms of scalability either well or extremely well. However, when it came to cost reduction the proportion of happy customers dropped sharply. Management complexity and interoperability between different cloud environments were also falling short of expectations.

Meeting expectations

Unpacking this gap between expectations and reality is tricky for enterprises because so many of the issues are linked. Cloud costs have crept up partly as a result of the complexity of hybrid/multi-cloud developments. The essence of the cloud is scalability but for many organisations that involves being able to scale up rather than down when less capacity is required. DevOps engineers can whistle up infrastructure in seconds but tend to be less efficient at decommissioning that infrastructure when it’s no longer required. If visibility of infrastructure is compromised by complexity – and it so often is – the chances are that there’s a fair amount of idle capacity which is still being paid for.

There is also the question of the commercial practices of cloud vendors and these are, quite understandably, designed to maximise customer spend rather than keeping it on its original budget trajectory. One of the primary drivers for a multi-cloud model is the desire to leverage cloud vendors commercially against one another and avoid becoming locked into a relationship with a single vendor.

Many organisations have benefited commercially by taking this approach and by the “race to the bottom,” pricing competition between hyperscalers for customers, but once those customers have been won, cloud vendors will deploy every sales and marketing tool at their disposal to scale up the services a customer uses and thus make it more difficult to shop around in future. Cloud vendors also tend to offer quite complex, variable pricing tariffs which almost seem to be designed to obscure clear visibility of costs.

Making cloud easy

The other, more technical, area where the reality of multi-cloud infrastructure is not quite matching expectations is in the interoperability and portability of workloads between clouds. Whilst containerisation should make applications more portable across different cloud environments, different CSPs usually support different APIs, data storage technologies, formats, and protocols.

Efforts have been underway for some time to resolve these issues and develop platform-agnostic APIs, containers and data standards, but progress has been slow and the shortage of cloud native skills is exacerbating the problem. This lack of standardisation and lack of skills is hampering the rearchitecting and migration of core business applications.

A purpose-built cloud management platform goes a long way to closing the gap between the theory and reality of cloud – and to finally enabling the migration of those core business applications currently keeping on premise datacentres on life support. Cloud platforms should ideally be capable of providing a holistic view of a heterogeneous cloud estate from a single viewpoint. This enhanced visibility reduces the risks of migration and simplifies management.

When workloads become truly portable between clouds, enterprises are free to truly leverage the best that each CSP can provide.