The following content was originally published here.
The initial wave of cloud migration has been about relocating workloads to cloud infrastructure. As business confidence in public cloud platforms increases, the pace of adoption has been accelerating with transformation projects happening across thousands of applications. The automated migration of workloads, primarily virtual machines, has been the dominant flavour of this wave, but very little of this transformation has affected the applications or data layers involved in these migrations.
This first wave has proved successful from the public cloud vendor perspective, with some clear winners gaining a significant share of the market. However, the next wave transformation is likely to be quite different in nature and will involve more application and data level changes. It is here that containers, APIs and Cloud Machine Learning are likely to be disruptive and promise to change the competitive landscape for public cloud vendors. The nature of competition is likely to shift from a pricing-centred IaaS approach to a feature-centred, value-driven approach around PaaS offerings. Below, we explore the reasons behind this shift.
A container-based approach
By enabling easy cross-cloud migration, containers are eliminating application level dependencies on public cloud features. Those who moved to cloud platforms early are now re-architecting their applications to leverage containers. This container-based microservices approach also means choosing the best fit for a particular microservice rather than a one-size-fits-all solution. Thanks to containers, it is now possible to easily port applications across environments including two competing cloud environments, disrupting the competitive landscape for public cloud providers.
Application and data layer services
Secondly, with several rich application and data layer services being hosted on public cloud, cloud is increasingly shifting its focus towards NoOps and serverless computing. Enterprises are realising the value of powerful PaaS offerings (for example, Google BigQuery or Amazon Lambda) for running special use cases, saving both time and effort. Similarly, several cloud-based APIs such as Google Maps, Speech, Translate and Text, offer capabilities that support business use cases and are otherwise difficult for a traditional enterprise to replicate. We are witnessing scenarios where a large part of a platform with commodity features is hosted on one public cloud and another part with premium features is on a separate public cloud, specifically to leverage its API ecosystem.
Finally, while there is a lot of general awareness about machine learning, enterprises are realizing that the technology is not easy to implement. Special algorithms take a long time to develop and perfect and this is the third area where clouds are starting to differentiate beyond basic IaaS play. Public cloud providers, either through in-house R&D or acquisition of boutique machine learning focused firms, are exposing larger suites of trained machine learning (and in some cases, deep learning) models through APIs for easy integration into cloud applications. Cloud platforms are also able to provide the required hardware (including GPUs where necessary) to deliver industry grade responsiveness from the trained machine learning algorithms. While initial launches have been around broad use cases like Google Speech, Vision, etc. it is likely that future launches will see algorithms for special industry use cases.
The cloud market is currently seeing strong price competition for commodity services in line with Moore's Law. However, it is logical to assume that cloud providers will want to start offering differentiated services to introduce more value-centric pricing, and then to retain workloads on the basis of these differentiated services. Containers, APIs and Cloud Machine Learning are set to transform the public cloud landscape significantly in the coming years.