Although the cloud has been with us for some time, many tech pros are still experimenting with the optimum way to make their tech stacks “cloudy.” Crafting a cloud-computing strategy comes with its own pitfalls, some obvious, others subtle but nonetheless destructive over time. Klaas Wijbrans, fellow architect for the Chief Architect Office at Philips, has quite a bit of experience with wrangling a variety of platforms, including the cloud, on-premises, in-product, and Internet of Things. Overall, he’s responsible for the company’s global reference architecture, leveraging 10 platforms. For those tech pros relatively new to the cloud, Wijbrans cautions that it demands a different way of thinking and programming. Concerns such as infrastructure, scalability, and hardware dependencies are no longer relevant. “Instead, people need to think about elasticity, security, and how to support agility,” he said. “For example, decomposing into microservices, with well-defined APIs enabling replacing complete functions with newer implementations.” Well-defined interfaces, he added, “allow more agility in the evolution of new functionality, different orchestration of microservices, and replacing implementation technologies.”

To the Cloud and Beyond

Once tech pros have built up some experience with the cloud, they may find themselves tasked with building out a company’s cloud-computing strategy. This is, obviously, an extraordinarily complicated endeavor, with many variables to take into account. For example, the cloud provides seemingly unlimited resources—which can lull tech pros into forgetting about the need for efficiency. An inefficient build will lead to additional costs once things are scaled up. Per Wijbrans: “One has to pay for what one uses.” Nor are all clouds the same, at least when it comes to project and company goals. An architect has to make choices based on a clearly defined mission, such as ensuring data portability (and security) across multiple “clouds.” If a company has an international reach, the architect must remain aware of how regulations regarding portability and security differ on a country-by-country basis. With data security, automation is vital, as many breaches stem from human error. “Deployment automation combined with security automation is one of the main ways to prevent this,” Wijbrans said, citing Philips’ investment in its healthcare-centric HealthSuite Cloud as an example of this. In addition, cloud vendors are constantly adding new services. “The architect needs to take this into account in his architecture,” Wijbrans said. “At the bottom of the architecture, he has to be able to adapt to the higher-level services, replacing proprietary implementations with the evolving commodity of the cloud vendors each time this is advantageous from a business or technical point of view.” (And without requiring a complete overhaul of the architecture, of course.)

Vendor Lock-In: A Real Fear

Whatever platform they choose, cloud architects and other tech pros may become paranoid about vendor lock-in. What happens if they want to move their data to another platform? Wijbrans believes that proper architecting can help mitigate their concerns. The real risk, he suggests, is the limited life cycles of products and services deployed to the cloud (as well as the lifespan of the accompanying data). “If the life cycle is short, lock-in should not be a serious concern as one can easily move the next product or service to another cloud vendor,” he said. “If the life cycle is long and the value and volume of data are high, this may be a serious issue: it is difficult and costly to move petabytes of data from one cloud to another.” In this context, “proper architecting” means not depending on a single vendor for data storage, if that’s possible. For some companies, it may also prove tempting to leave data on-premises, but cloud providers have become much more sophisticated with regard to data security and availability (and cheaper in terms of storage costs). Because Philips focuses on healthcare, Wijbrans sees his architecture work as an opportunity to help improve lives. A cloud-based approach to healthcare means that products and services that once existed in their own “silos” can now be integrated as part of larger, more holistic solutions, while adhering to more strict privacy requirements. In turn, that leads to platforms such as the Philips Clinical Analytics Platform, which can deploy and leverage deep learning (as developed on its Insights platform) to detect new patterns in a patient’s condition.