In a previous blog post, We’re Collecting Valuable Data. Now What?, I discussed how the combination of a data-driven culture and using cloud infrastructure can help companies aggregate and analyze data that they collect across multiple sources inside and outside the organization. Now I’d like to take a closer look at how a multicloud environment makes this process happen.

A disciplined multicloud strategy, including a hybrid architecture of on-premises and cloud, provides a flexible and secure way to pull in data from internal sources. This strategy can include transactions, website activity, device sensors, and external factors, including social media, government data, and many others. More importantly, it offers access to best-of-breed data management and analysis solutions to transform your data into actionable knowledge in real time. Multicloud is not without its caveats, which I examine as well.

The role of shadow IT

For many companies, the road to multicloud has been a winding path instead of a strategic imperative. As business users across the enterprise adopt cloud-based point solutions without going through the normal procurement process, there has been an advent of so-called “shadow IT.” This rise leads to questions: How can we transform this unplanned adoption of multicloud into a viable business strategy? How can we make sense of the chaos?

An organization can take one of three paths.

Single vendor

The simplest strategy is adopting a single vendor policy. All your data and your workloads go into one cloud. With this approach, you don’t need to constantly reevaluate data source location, and you don’t need to move data around, which is not an ideal situation. This approach means that you’re betting on one vendor, so you should be certain of that vendor’s reputation, longevity, and commitment to providing a valued partnership, as well as ongoing cost savings. Also consider vendors that have vertical integration between their SaaS enterprise and industry-specific software and cloud infrastructure. This strategy minimizes the need to move data.

If you go with a single vendor, be aware of the risk of vendor lock-in. To minimize it, use cloud standards, such as Kubernetes, and avoid the lure of using proprietary vendor code or other features that could make migration harder, should you ever decide to move.

Multicloud

Instead of requiring business users to adhere to a single-vendor policy without exception, a more flexible approach is to embrace multicloud by distributing your data and workloads among different vendor clouds. This approach adds complexity and requires deep strategic thinking about the part each vendor plays. What are their individual strengths and weaknesses? How well can they share data?

Not all cloud providers play well together. Look for a vendor that offers native multicloud connection capabilities to avoid frustration, lock-in, and higher costs down the line. It’s also essential to consider the bandwidth cost of moving data between regions or in and out of the cloud, which can be substantial.

To make a multicloud approach work, you need a monitoring and managing layer. Otherwise, you’re wasting time and energy with point-to-point integrations and moving your data and your workloads from vendor to vendor, resulting in diminished returns. Initiatives like the Cloud Native Computing Foundation and Kubernetes, for example, provide a programming infrastructure and paradigm to manage this increased complexity.

Hybrid cloud

The third option is an extension of multicloud. With this architecture, you’re mixing and matching data and workloads among multiple clouds while also adding on-premises infrastructure or private cloud. This is the most complex option, but also the most flexible one. Organizations can use their existing infrastructure efficiently while integrating it with cloud solutions.

To optimize a hybrid cloud architecture, look for cloud providers that already have a deep footprint in existing applications and data and the capacity to integrate with them as seamlessly as possible. Hybrid cloud facilitates an incremental evolution toward digital transformation, allowing organizations to transform at their own pace. It also accommodates the simple reality that, depending on your industry and your needs, some workloads and some data are probably always best suited for on-premises, whether it’s to meet performance or regulatory requirements.

Applying advanced technologies

In addition to providing access to multiple disparate forms of data for analysis, a multicloud approach gives you a veritable buffet of solutions that use artificial intelligence (AI) and machine learning (ML) to automate processes and deliver fresh insights. But there’s always a caveat when talking about the latest technology trend. In our excitement, we often forget that technology is a tool, not an end in itself.

The first rule of data-driven application development is that you shouldn’t really approach things as, “Let’s build an AI or ML application.” Instead, solve a problem or create an opportunity using the right technology for the job. Start with the basics. If you want to automate a process, for example, begin by analyzing the process itself, and then give the people working on the process the right software tools. The fact that many cloud solutions offer out-of-the-box AI or ML capabilities that don’t require specialized programming can speed up the process.

What to take away

The final thought I want to leave you with is that there’s no one-size-fits-all solution. That’s exactly the point of a multicloud approach. A startup with no IT infrastructure but a deep knowledge of cloud options is in a different position than an enterprise with an existing investment in its own data centers but minimal expertise in working with multiple cloud vendors. So, your mileage varies depending on where you stand. It’s a journey, and I invite you to learn as you go along.

Oracle Generation 2 Cloud Infrastructure can power your hybrid multicloud strategy. To learn more, visit Oracle Cloud Infrastructure.

George Anadiotis is the founder of Linked Data Orchestration, where he works on the intersection of technology, media, and data.