The Clouds Roll In

Like houses, cloud architecture is a matter of function, need and money, with no one-size-fits-all strategy. Consider your many options.
Share on facebook
Share on twitter
Share on linkedin
Share on email

Ask three IT leaders in different industry sectors for their cloud strategies and prepare for different answers. Like its meteorological namesake, the cloud is a mix of different types, requiring complex decisions on the right combination.

At Varo Bank, for instance, Chief Technology Officer Deep Varma’s deliberations have resulted in a multi-cloud environment involving Amazon Web Services’ Infrastructure-as-a-Service cloud offering, bank transaction processing provider Temenos’ Software-as a-Service (SaaS) cloud solution, Visa’s SaaS Cloud Connect debit card solution and a variety of cloud plug-and-play applications.

Varo Bank CTO Deep Varma

“Our cloud architecture is built to allow decoupling, where each computing component is independent and autonomous until they need to come together to execute a task,” said Varma, noting that the all-digital bank, the country’s first to receive a national bank charter, depends on the cloud strategy to maintain 24/7 uninterrupted operations, a key plank in its value proposition.

Unlike six-year Varo Bank, which has no on-premises servers and computers, CNA Insurance, a 124-year-old provider of property and casualty business insurance, has substantial on-prem computing assets, courtesy of its venerable age. The public company touts SaaS cloud providers like Microsoft Office 365 and Workday (for HR and financial management tasks), whose applications integrate with its on-prem computers and servers. But the insurer’s primary cloud provider is Google Cloud Platform (GCP), a Platform-as-a-Service (PaaS) provider with more than two-dozen cognitive computing tools like Vertex AI, which CNA uses to create models, such as those designed to help price insurance and manage claims fraud.

“We’ve got a hybrid cloud strategy but maybe not a `classical hybrid,’ given the SaaS multi-cloud aspects,” said Jane Possell, the company’s Senior Vice President and CIO. A hybrid cloud environment, unlike a multi-cloud one, refers to a public cloud like GCP along with an on-prem computing infrastructure, one that Possell said she is working with Google to migrate into the GCP platform.

SEI, a 53-year-old large wealth and investment management solutions company, also has substantial on-prem computers and servers. The public company is pursuing a cloud native strategy, an architecture in which applications are designed for the public cloud but can be deployed both on SEI’s on-prem computers and servers as well as the public cloud.

“We made an aggressive shift about two years ago to designing and developing software with cloud native principles, giving us the opportunity to rapidly design and build new applications and, when needed, connect them using APIs (application program interfaces), giving us the option to deploy them either on public cloud or on prem,” said Executive Vice President and CIO Ryan Hicke, referring to SEI’s cloud approach as a multi-cloud strategy.

Obviously, there is no standard cloud adoption model, with many other CIOs pursuing equally unique approaches in designing their companies’ cloud architecture—hence broad terms like “hybrid,” “multi-cloud” and “native.” “Ask a CIO where they are today in their cloud strategy, and you will get varying answers,” said Nicholas Merizzi, a Deloitte principal in charge of the consulting firm’s cloud strategy services.

Due to the complexities in formulating an optimal cloud strategy, these answers are hard to come by. “Among CIOs, the decision is akin to a religious debate, with starkly different points of view,” said Prasad Sankaran, a partner at Bain and Company and a member of the consulting firm’s global Enterprise Technology practice. “The goal is to make a decision that makes sense for that organization, based on its core IT systems first and foremost. Once decided, however, the result is often a cloud strategy that is sui generis—inherently one-of-a-kind.”

I Did It My Way

Five years ago, many midsized and even larger companies were still making their way to the cloud. No longer is this the case, said Merizzi, who speaks almost daily with CIOs across different industry segments to glean their cloud adoption trends. “If you look across the Fortune 250, almost every one of those companies has a cloud-first strategy in place, albeit at different stages of what is essentially a long journey,” he said. 

This journey is a twisted trek with a fork in the road—maintaining status quo on-prem computing environments or migrating applications and data to still-evolving cloud models. Despite an accelerating shift to the public cloud, up some 23 percent in 2021, according to Gartner, most companies’ data remains spread across on-premise computers and servers and public cloud infrastructures.

“Big iron-type mainframe systems in many large enterprises are 40 years old in some cases, which is a barrier impeding a more robust migration to a cloud architecture,” said Merizzi. “Mid-market companies have a unique advantage, as their smaller IT budgets can access advanced cloud services, without the tough decision big companies have to make. The adoption rate is unfolding faster in this space.”

There’s much to choose from in building this architecture. A hybrid-cloud structure generally is composed of on-prem servers and computers, private cloud services and a Platform-as-a-Service (PaaS) public cloud like AWS, GCP or Microsoft Azure providing wide-ranging cloud resources on a pay-as-you-go subscription basis, including enterprise applications and cognitive computing tools. A multi-cloud structure typically includes a PaaS public cloud provider, a private cloud for the company’s systems of record, and one or more Infrastructure-as-a-Service (IaaS) public cloud providers offering compute, storage and networking resources on demand. 

While many enterprises have migrated their data and operations to the public cloud, some applications continue to work just as well and even better on on-prem computers and servers. “My view is that there is no reason to close down an on-prem data center and do a `lift and shift’ to the cloud, unless there is some sort of financial need to do it,” said Sankaran.

“Your company’s history is in that data center,” he said. “The truth is your legacy applications are not going to function any better in the cloud, which is just a cheaper data center. You can persist with those apps on-prem until reaching a point where you can replace them with other apps in the cloud.”

His comments do not downplay the extensive capabilities of public cloud providers. GCP, AWS and Microsoft Azure are positioned to design, develop, manage and orchestrate a company’s data, software and applications. A particular allure is their query-driven cognitive computing tools, which are accessible to everyday users of information systems and not just highly technical data scientists. GCP, for example, offers a suite of tools that includes advanced data analytics, AI and machine learning, 3D visualization, custom models that detect emotion, natural language processing (NLP) and a fully managed environment to run containerized applications. That’s the short list.

Small wonder CIOs are bolting together interoperable hybrid clouds and multi-clouds. “All our clients, by and large, are multi-cloud, with at least half a dozen or more on a SaaS basis, plugging into new AI and machine learning modules to create and deliver differentiating experiences,” Merizzi said.

Sankaran agreed that an optimal cloud strategy can reap a competitive advantage. “For every company and particularly mid-sized ones, the cloud is an imperative, but the challenge is developing the right architecture and strategy without painting yourself into a corner,” he said. “There is no standard cloud adoption model, which has led some CIOs down misguided paths. Determining whether to maintain systems and applications on-premises as cloud native or moving and locking them into public clouds is not a trivial decision, which is why it makes sense to have advisory help to guide a clearer path.” 

Mix and Match

At CNA Insurance, SEI and Varo Bank, their varying cloud strategies are hinged to a technology roadmap guiding IT in support of the organization’s long-term strategic plans. At Varo, the bank’s strategy is to rapidly grow its customer base, a goal it is reaching through highly competitive interest rates and instant cash advances, which it attributes to a low-cost all-digital operating model. The bank’s customer accounts reached 4 million in September, up from 2 million in August 2020 when Varo received its bank charter. Over the 13-month period, its valuation tripled from $700 million to $2.5 billion.

Varma has been instrumental in building Varo’s IT stack, which allows for services that are more agile and more personalized than what traditional banks offer, he said. Cloud is a big part of this architecture. “We built our technology with one mindset—to be able to decouple from AWS as our platform provider and `plug and play’ elsewhere, if so desired,” he said.

Within this multi-cloud architecture, Varo’s front office customer-centric applications reside in AWS, which manages the physical cloud infrastructure but not the decoupled cloud applications hosted on it. Specific work can then be guided to a SaaS cloud application like Temenos, a self-service portal offering users end-to-end retail and corporate banking services. “We get the benefit of their technology solutions, which we can tie back into our core banking system using AWS’ Direct Connect dedicated network connection solution,” said Varma. 

CNA’s hybrid cloud strategy accounts in part for the insurer’s significant on-prem computing resources and data, which Possell is migrating to GCP. She estimated that 10 percent of applications are presently running in GCP’s cloud, which will accelerate to about 50 percent to 60 percent over the next three years. “Twenty percent may never go into Google Cloud, such as the Workday, Microsoft Office 365 and Guidewire applications, which are on-prem (on a SaaS cloud basis),” the CIO said.

In the meantime, the company is leveraging GCP’s estimable data and analytics toolsets. “There’s no question that GCP has impressive use cases in data and information, given their roots as a search engine, as well as fantastic data management capabilities,” she said.

She provided the example of Vertex AI, GCP’s unified machine learning platform, which is touted for making it easier and faster for everyday users to build, deploy and scale machine learning models. Minimal expertise is required to build the models, which involve nearly 80 percent fewer lines of coding. “We’re GCP’s alpha customer in the insurance sector (using the tool) and beta in the sense that some large auto manufacturers have previously used it,” said Possell, adding that the insurer is creating models using the data query feature much faster than it could if the data were stored in multiple clouds.

SEI is on a very different journey, tapping the investment management firm’s IT skill sets to design, build and use new cloud native applications, optimize current on-prem applications, connect them all using APIs and then run the apps on its proprietary platform, which includes third party public cloud providers. “Instead of saying let’s lift and shift everything we do to a provider like AWS or GCP, we look for applications that would specifically benefit from being in the public cloud,” Hicke said.

He provided the example of using third party machine learning tools in front office customer-facing administrative functions to increase efficiency through analytical insights. SEI also integrates SaaS cloud-based single product solutions like ServiceNow to manage workflows. SEI is not an outlier, with many companies also opting for a cloud native architecture—not with the intent of growing their on-prem footprint but rather to harness the capabilities of the public cloud on an as-needed basis.

Down the line, expect continuing iterations in these cloud adoption trends and models. Much like Amazon launched with a compelling online alternative to buying books, before selling everything under the sun on the internet, AWS and the other PaaS cloud vendors are building platforms that will be tough to ignore, said Merizzi.

“If you look across GCP, AWS and Microsoft Azure, right now they offer about 150 different `Lego blocks,’ a number I’m making up but is probably pretty close,” he said. “Five years from now, there will be exponentially more of these `Lego blocks’ for users to pick and choose from in building and delivering differentiated experiences, with the goal of carving a sharper competitive edge.”

Leadership Advice for Today’s Strategic CIO

Subscribe to our Weekly Newsletter

MORE INSIGHTS

Leadership Advice for Today’s Strategic CIO

Subscribe to our Weekly Newsletter

Leadership Advice for Today’s Strategic CIO

Subscribe to our Weekly Newsletter

Scroll to top