Does AI-driven cloud migration herald the death of re-hosting?

This article first appeared in Light Reading.

In the midst of their cloud migration journeys, many communications services providers (CSPs) are finding themselves underwhelmed by the outcomes to date.

Comprising thousands of applications and workloads, migration for these players is so complex, costly and time consuming that, with a few notable exceptions, mass migration simply hasn’t happened. Yet, change is on the cards with the integration of automation and artificial intelligence (AI) capabilities with modern migration tooling.

The challenges of mass migration

Already, leading CSPs have largely embraced cloud-first strategies, focusing on developing new software and assets designed for the cloud, referred to as ‘net new development.’ However, migrating their existing assets to the cloud remains a challenge, as the scope of most migration projects is relatively small in the context of a large enterprise, typically encompassing tens or perhaps hundreds of workloads. Furthermore, evaluating these workloads to determine suitability for cloud migration requires knowledge of the individual applications as well as cloud technology expertise. Moreover, much of the time, the process relies on gut instinct and subjective opinion. The bottom line is that there is no standardized playbook to determine the ideal migration path.

Duration too presents a challenge. Current estimates and experience show it takes at least five to seven years for operators to move its IT estate to the cloud. In a bid to accelerate the process, workloads are often re-hosted using the lift-and-shift method. But this seriously limits the benefits of migration as applications aren’t primed to take advantage of cloud capabilities such as auto-scaling. What’s more, it may become more expensive to run these applications in the cloud, and harder to manage them.

While there are ways around this (when workloads are modernized via re-platforming or re-factoring, cloud benefits can be achieved), this requires significant investment and specialist cloud skills. Automating migration can help too, but there is still a lot of manual work involved.

Yet as things stand, despite their cloud-first ethos, most CSPs still have a large on-premises estate, leaving missed cloud business goals, and disillusioned telco leaders and managers.

AI spells the death of re-hosting

AI-driven tools that are beginning to appear on the market will soon have the power to “flip the script”, delivering a fast, highly automated end-to-end process for cloud migration. Specifically, they will enable ‘modernization-first’ mass migration at a reasonable price point, allowing operators to realize cloud benefits at scale. So instead of re-hosting being the default option, it will only be necessary for a minority of applications.

How will this work in practice?

We envisage a new age of automated, intelligent migration, where AI-driven tooling conducts an automated scan of each workload before recommending the most appropriate ‘R’ (re-host, re-platform, refactor etc.). Tools will then generate an estimation of effort and resources required to execute the migration. Further recommendations might focus on the prioritization of workloads in groupings aligned with business processes and the construction of optimized waves of migration. Furthermore, AI tools could propose ways to break down legacy code into microservices and automate the design of containers.

We also see AI tools having the ability to identify technical debt, empowering engineers to fix problems ahead of migration or implement automated fixes. In addition, AI will execute code adaptations, integration with third-party tooling like ticketing systems and more.

Significantly, all of this will happen in an iterative manner, with AI software scanning and rescanning applications repeatedly, to continually suggest and execute further improvements. Over time, migration activities will increasingly be handled by the software itself – even the coding.

Modernization-first will deliver more cloud value

Ultimately, as AI-driven approaches gain momentum the demise of re-hosting indeed seems imminent. If today, roughly 80 percent of workloads are re-hosted and 20 percent modernized, in the future, 80 percent of applications will be modernized. And with more workloads moving to the cloud – and performing better there – a virtuous circle will arise where value and business returns stimulate further cloud adoption. The end-to-end process will also become quicker, cheaper and more predictable, resulting in a stronger business case for moving to the cloud.

In this new paradigm, the baseline term for this process will evolve from ‘cloud migration’ to ‘cloud modernization’, with AI tools unlocking the boundless potential of cloud adoption for CSPs and other large enterprises alike.