Optimizing Cloud Solutions for Government

The rapid evolution of cloud solutions is laying to rest many of the old build-vs.-buy deliberations IT leaders once had to make. “Those types of decisions are still there, but cloud is becoming much more viable for a larger set of the sorts of applications that governments run today,” says Center for Digital Government (CDG) Senior Fellow Otto Doll. At the same time, that question is being replaced by a new one, Doll says: “Cloud gets sold as this turnkey process, and it can be … [but] we are challenged by whether we are truly getting the benefit from being in the cloud that everyone loves to tout.”

Much of the discussion about optimizing cloud operations involves developing systems to track workloads, usage and costs over time. This can be especially critical when demand surges, as it did when governments shifted many business functions online during the COVID-19 pandemic. However, the design and implementation of cloud applications can have an even greater impact on their overall efficiency and ultimate success. Among the factors to consider:

Assessing not just one application at a time, but the impact of managing multiple cloud environments. Maintaining all applications and data in a single cloud environment “is too restrictive for any jurisdiction,” says Doll, who previously served as CIO for the city of Minneapolis. As a result, governments will be managing hybrid environments for the foreseeable future — multiple clouds hosting different applications and data operated by various service providers, as well as connections with on-premises data centers and applications.

“I’ve got to observe what’s going on across all my clouds, and the interactions among the clouds,” Doll says. When jurisdictions decide to pull together data from different departments or applications to address new business problems, for example, “you need to have the clouds talk to other clouds to transact business without humans being involved,” Doll says. Otherwise, “you can run into nasty scenarios where you can’t get here from there,” he adds.

Determining the level of integration throughout the enterprise. Ensuring integrations among different cloud environments and systems are effective and scalable is critical to managing hybrid multi-cloud environments. “If you don’t have a true understanding of the whole system — from the applications layer to the infrastructure layer and everything in between — in a multi-cloud environment, then it isn’t optimized to maximize your technologists’ ability to administer that world,” Doll cautions.

Identifying opportunities for automation. Understanding the extent to which you can automate cloud systems and integrations is critical to determine IT's ability to manage new systems and spin up additional resources as demands shift.

Understanding what data you need — and in what timeframes. Governments need to ensure cloud applications are optimized to send and receive data at the speed of business. For example, in Minneapolis officials wanted to download data from a cloud-based 311 system into the city’s data lake every 15 minutes. The information was needed to continuously update a business intelligence platform. “What happens if I needed to update that in real time every five seconds instead of every 15 minutes?” Doll asks. “Optimizing your ability to do that [involves] having your cloud providers buy into that up front.” You may also need to spell out these provisions in contracts and service agreements.

Effective data interchange becomes even more critical as the Internet of Things (IoT) and all the associated cameras, sensors and other devices integrate with cloud solutions. “All these devices are collecting massive amounts of information and streaming it in real time,” Doll says. “If I’m monitoring traffic, it doesn’t do me any good to know an hour after the fact.”

Verifying the ability to scale applications. While governments may be able to spin up a cloud environment based on shifts in demand, compute and bandwidth aren’t the only factors. Can applications hosted in the cloud keep up with these shifts if traffic increases dramatically, as it did in many cases during the pandemic? Will the security around the application and endpoints keep pace to ensure constituent data remains safe?

Developing systems to monitor and optimize ongoing costs. IT staff must find ways to track and model usage on an ongoing basis, including identifying idle or inactive systems and assessing whether they are assigning the right workloads to the right cloud resources at the right scale.

Considering opportunities to apply artificial intelligence or machine learning (AI/ML) to operations. Doing so can minimize administrative tasks around optimizing cloud usage that IT staff would otherwise be manually managing.

As was the case with previous technologies, many of these factors may not be visible to the end users of cloud solutions, but they are essential for those who must manage them. In a way, that’s a sign that the technology has become a mainstream part of enterprise IT, Doll says. “The stuff behind the curtain that folks outside of the IT organization never see tells me that cloud has come of age,” he says.