F5’s Lori MacVittie on Superclouds, Hybrid IT and the Future of Cloud

Cloud infrastructure is becoming a set of resources that organizations combine and move among, rather than a permanent home for every application. That shift is the idea behind the supercloud: an architecture intended to make applications work across different providers, company data centers and edge environments through a more consistent operating layer.

Lori MacVittie, a distinguished engineer and chief evangelist in F5’s office of the CTO, describes a supercloud as an architecture that lets hybrid IT operate across every layer of the technology stack. The practical aim is to make deployment and day-to-day operation less dependent on the particular tools and interfaces of each environment. MacVittie has more than 25 years of experience in application development, IT architecture, and network and systems operations. She joined F5 in 2006.

Networking is the connective layer

For MacVittie, a supercloud depends on solving a basic problem: applications that span environments need to communicate across networks built with different technologies and interfaces. Multi-cloud networking can help by acting as a connective layer between data centers, public cloud services and edge locations.

That layer can make connectivity more consistent from network layer 4, which handles transport, through layer 7, where application traffic is managed. Standardizing those functions across environments also creates points where organizations can apply security controls. Without a common approach to networking, a single interface for managing workloads across clouds would be difficult to deliver.

The idea is not that every provider becomes identical. Each environment still has its own services and tools. Instead, the supercloud concept aims to make those differences easier to manage, so teams can choose where to run an application based on requirements such as cost, performance, scale or sustainability.

Hybrid and distributed applications are already common

MacVittie characterizes hybrid computing as the norm for enterprises. Research she cited found that 88% of enterprises operate in a hybrid state, and a similar share manage application portfolios that combine traditional and modern software. She also expects hybrid deployments to persist as organizations plan to run artificial intelligence workloads both on premises and in public clouds.

Many applications are distributed across locations already. MacVittie said 60% of enterprises in the research operate hybrid, distributed applications, with components of a single application running in different places. As companies assemble digital services from multiple applications and services, those components may sit in a company data center, a public cloud, at the edge or in software as a service. APIs connect the pieces.

That structure gives teams more options for placing workloads, but it also increases the number of connections and operating contexts they must manage. A common networking layer can help, though organizations still need to account for each service’s specific requirements and maintain visibility across the whole application.

Workload placement can include sustainability

Being able to move workloads between environments may let enterprises consider environmental impact alongside performance, capacity and cost. MacVittie calls this the agility cloud computing was expected to provide, a capability she says is only now becoming practical for many organizations.

She also pointed to public cloud providers’ commitments to measure and report environmental impacts, and to offer services that help customers make choices aligned with their own goals. The usefulness of these options depends on whether enterprises can compare them and incorporate them into workload decisions. The broader supercloud idea could make placement more flexible, but the source does not specify a common measurement method for weighing environmental impact across providers.

Security and data quality remain operational concerns

Cloud security has long involved shared responsibility. Providers protect their infrastructure and services, while enterprises remain responsible for securing the workloads they run. MacVittie particularly stresses using providers’ available tools to protect data stores, which have been a common route for data exposure.

In her view, the growing use of AI is drawing more attention to cloud security and compliance than cloud adoption alone. Cloud providers are major suppliers of AI services, adding another reason for organizations to examine how those systems and their data are protected.

Data quality, she argues, is not inherently a cloud problem or an on-premises problem. It depends on how an organization collects, processes and manages information. AI makes the task harder because it relies on data and produces data of its own. Cloud services may offer tools to monitor and improve the quality of both inputs and outputs, but teams still need sound data practices and pipelines.

Cloud economics may change as usage shifts

MacVittie sees the cloud’s next major change in its business model. Providers have traditionally charged for a range of resources and services, while enterprises have become increasingly attentive to costs. Cloud spending can be difficult to forecast, which makes it harder to plan workloads and weigh cloud use against other priorities.

She expects providers to adjust billing in ways that make costs easier to predict and could encourage customers to use more services when they can understand the expense. That change matters as organizations treat cloud capacity as something they can call on temporarily, rather than a permanent destination for every workload.

MacVittie says cloud is now largely commoditized. Enterprises continue to move workloads into and out of cloud environments, and increasingly operate their own infrastructure with practices associated with cloud providers, including site reliability engineering. Providers may need to make their resources easier to consume on demand, rather than assume customers will settle permanently into a single environment.

Serverless and edge extend the options

Serverless computing pushes the pay-as-you-go model further: teams don’t provision a server in advance. Instead, the resources needed to run code are made available when the code is invoked. This can reduce costs tied to idle capacity, though the source offers no figures for savings or guidance on which workloads benefit most.

MacVittie expects edge computing to remain significant. Internet of Things systems already use edge resources, and organizations’ interest in managing where data is handled has increased attention to edge data services. Edge providers are also building distributed platforms for security and delivery that can serve organizations with hybrid and multi-cloud applications.

She expects edge’s role to grow as AI changes how applications are built and how people interact with technology. Vertical clouds, by contrast, may have limited room to expand: MacVittie’s view is that most industries don’t need specialized cloud environments, and industries that do have largely established them.

The practical test for the supercloud will be whether it makes distributed applications easier to run while giving teams useful control over placement, networking, security and cost. For organizations already combining traditional systems, public clouds and edge services, the value will depend on how well that common operating layer handles the differences between them.