Many organizations end up with servers in a dozen places, from a head-office server room to stores, plants and cloud accounts, each run a different way. Distributed hybrid infrastructure (DHI) is a label, used by the analyst firm Gartner among others, for products that tackle this by treating that scattered hardware as one system. Compute, storage and networking sit wherever the workloads need them, but teams provision, update and monitor them through a single management and control plane, with the self-service and automation habits of public cloud. Because the term names a market grouping rather than a standard, its edges can move, and vendors often sell similar products under other names.
At a glance
- DHI is a market category, not a technical standard; the term comes from Gartner, and other firms and vendors use different labels for similar offerings.
- Its defining traits are infrastructure placed in many locations, one management and control plane over all of it, and cloud-style operations such as self-service and automation.
- Offerings come from public cloud providers extending into customer sites and from infrastructure vendors adding cloud-style management to on-premises systems.
- Common drivers are data residency, latency, local processing at the edge and applications that can’t easily move.
- Capabilities, connectivity needs and commercial models, from outright purchase to subscription, vary widely between offerings.
What problem it solves
Many organizations want public cloud’s way of working, with fast provisioning, automation and a single management console, but cannot put every workload in a public cloud region. Some data must stay in a particular country or building. Some applications need very low latency to machines on a factory floor or in a store. Some systems are too costly or risky to rewrite. The result is often separate environments, each with its own tools, skills and contracts.
DHI products aim to narrow that gap. They let an organization run infrastructure in its own data center, a colocation site or an edge location while managing it in a way that resembles, and is often connected to, public cloud. That can reduce the number of separate operating models and make a hybrid cloud strategy easier to run.
How it works
Products in this category differ, but most share a few elements.
Infrastructure at the customer’s location. Servers, storage and networking, sometimes supplied by the provider as an integrated system and sometimes running on hardware the customer chooses, often in a hyperconverged infrastructure (HCI) design.
A cloud-style control layer. Software that provides self-service provisioning, automation through APIs and centralized management, so teams can deploy workloads in a similar way across locations.
Connection to a cloud. Many offerings are managed from a public cloud console and extend some cloud services to the local site. How much works when the connection drops varies by product.
Commercial model. This is not part of what makes something DHI, and it varies: some offerings are bought outright, others are sold by subscription or by usage. Terms, minimum commitments and hardware ownership differ.
Shared operations. Depending on the offering, the provider may handle hardware maintenance, updates and monitoring, while the customer handles the facility, power, local network and workloads.
When it matters for buyers
- When some workloads must stay on premises or in-country. DHI is one way to meet residency rules while using cloud-style tools; sovereign cloud offerings are another.
- When edge sites need local processing. Retail, manufacturing and healthcare often run applications close to the people and machines they serve.
- When refreshing data center hardware. It is worth comparing a conventional refresh with a cloud-managed or subscription option.
- When standardizing on one cloud provider. Extending that provider’s platform on site may simplify skills and tools, at the cost of deeper dependence.
To compare on-premises and private cloud options, see our private cloud solutions overview.
Questions to ask vendors
- Which services run locally, and which still depend on your public cloud?
- What happens to running workloads if the connection to your cloud is lost?
- Who owns the hardware, who maintains it, and what are the response times?
- What are the minimum commitments, term lengths and costs to exit?
- How are updates applied, and can we control their timing?
- Which compliance certifications cover the on-site deployment, not just your cloud regions?
- What facility, power and network requirements must our site meet?
How it differs from hybrid cloud
Hybrid cloud is a deployment model: an organization uses public cloud alongside private or on-premises infrastructure. Distributed hybrid infrastructure is an analyst label for a type of product that helps run that model consistently, by managing infrastructure in many locations through one cloud-style control plane. A company can run a hybrid cloud with separate tools for each environment and never buy a DHI product; DHI is one way to make hybrid operations more uniform, often by tying them more closely to a single provider such as a hyperscaler.