A composable infrastructure is an IT framework for delivering compute, storage and network resources as a set of...
unified services that could be the next evolution in hyper-converged infrastructure.
Composable infrastructure addresses many of the limitations of hyper-converged systems; it offers a more flexible, software-driven choice for quickly and efficiently accommodating changing workloads. Because of this flexibility, this new framework may represent the next generation of data center architectures.
Confusion often surrounds the meaning of composable infrastructure, in part because that term is often used by vendors interchangeably with labels such as software-defined infrastructure (SDI) and infrastructure as code (IaC). One way to think of a composable infrastructure is as a type of SDI that disaggregates and logically pools hardware resources and treats each one as a service that IT can provision on demand and tailor to specific workloads.
This disaggregated infrastructure uses software-defined intelligence to pool and control the resources via a comprehensive API. Together, the software and API make it possible for you to allocate resource pools or reallocate -- compose or recompose -- as needed to support different application requirements. This provides flexibility similar to that of a public cloud service.
An IT team can implement a composable infrastructure on bare metal, virtual machines or containers. The team can start small, with as little as a half-rack of hardware, and then add equipment incrementally, according to application requirements. This new type of infrastructure can also support IaC, so developers can add code to their applications to specify the resources needed to run those applications without administrator intervention.
How it works
Composable infrastructure relies on three technology layers to deliver a unified framework: the hardware resources, the intelligent software and a management API.
The hardware layer includes the components that make up the compute, storage and network resources. Ideally, the hardware can span multiple locations and include any disaggregated, industry-standard components, although today's composable infrastructure products have not yet achieved this level of flexibility.
The software layer abstracts the hardware resources and organizes them into logical resource pools. The software is both programmable and template-driven, and it can provision services, automate operations, self-correct resource pools and perform numerous other tasks.
The software relies on the API layer to communicate with and control the hardware resources. The API should be open and extensible enough to interact with various types of hardware and to facilitate platform integration with third-party systems and services, such as reporting or management tools. A comprehensive API, like the intelligent software, is essential to the success of the new architecture.
Benefits of a disaggregated infrastructure
Although it is similar to hyper-converged infrastructure (HCI) in some ways, composable infrastructure offers much more flexibility. The typical HCI is preconfigured for specific workloads, and scalability is limited to a strictly defined trajectory. The HCI system also tends toward vendor lock-in, especially when IT buyers purchase a preconfigured system.
This latest type of infrastructure can adapt to changing workloads with relative ease, which enables IT teams to quickly allocate the resources they need when they need them. A disaggregated infrastructure should also be able to accommodate hardware from different vendors, although, in practice, vendor lock-in can still be an issue.
Even so, this new infrastructure framework enables IT teams to better use hardware resources, which can lead to significant cost savings. In addition, because of the intelligent software, IT needs fewer resources to manage the environment and deploy applications, resulting in more streamlined operations and additional savings.
IT teams can also scale up the hardware incrementally in a disaggregated infrastructure environment to meet the demands of the users. This means that IT admins no longer need to over-provision resources based on anticipated workloads.
Despite the advantages of composable infrastructure, one of its biggest challenges comes from the lack of industry standards for how to deploy it. This has led vendors such as HPE, Dell and Liqid to each take a slightly different approach to their product lines, resulting in proprietary products and the risk of vendor lock-in.
A true composable infrastructure should be open and extensible, and it should support hardware from multiple vendors. But this is still a relatively new industry, so it's no surprise that standards have yet to emerge.
In addition, the software that drives a disaggregated infrastructure is still maturing, and it could be a while before we get this newest architecture in the truest sense, in which any standards-based components can be used and all those components are disaggregated.
Composable infrastructure is off to a strong start and is showing plenty of promise. Before long, it could have a strong foothold in the data center, much like the converged and hyper-converged systems implemented in recent years. The question then will be how to use all those old systems once the new ones have taken hold.
Learn composable infrastructure's pluses and minuses
Promise of composable infrastructure catches IT admins' eyes
Composable infrastructure blazes trail for infrastructure as code