Today’s modern data centers are among the most complex and technologically advanced locations on the planet. They sport extremely powerful server and network systems that allow customers to have powerful resources at their disposal for running production workloads.
Over the years, the various technologies and methodologies used in data centers have been changing and advancing. Today’s customers consuming data center resources have some of the most complicated and varied requirements that include a wide number of different types of technologies and resources. This may span everything from physical bare metal servers, virtual machines, containers, and other hybrid cloud-driven applications.
As datacenter technologies have advanced and software-driven solutions have become more powerful, the datacenter has given rise to a new type of infrastructure – composable infrastructure.
What is composable infrastructure and how does it compare to enterprise datacenter technologies that are commonly used by enterprise customers today?
Public Cloud’s Link to Composable Infrastructure
Public cloud has arguably had a greater impact on enterprise data center technology, infrastructure, and services than any other development in recent years. The public cloud revolution has literally reshaped the way that enterprise customers think about provisioning infrastructure and services for customers and employees. Since the mass adoption of public cloud, organizations are now able to shift major capital expenditures into much more palatable operational expenditures for infrastructure and services.
However, perhaps the more important set of changes that have been instigated by public cloud adoption is the way that organizations can now handle provisioning and maintaining infrastructure. Since public cloud infrastructure and resources can be interacted with via APIs provided by the cloud vendor, infrastructure and infrastructure services can now be controlled, manipulated, provisioned, configured, and disposed of via code. This has precipitated the whole “infrastructure as code” movement. This new methodology and “mindset”, facilitated by massive public cloud adoption, has also changed the way that enterprise customers think about any infrastructure, even infrastructure on-premises or in colocation data centers.
Public cloud introduces a new way of dealing with resources such as compute, memory, network, and storage in a way that deals with these in “pools” of resources that are utilized and can be given back to the specific resource pool. This is a much different approach to traditional infrastructure which is typically provisioned out of the box for a specific application or purpose and those resources are never repurposed or reclaimed. The same holds true for virtual machines typically.
Why is On-Premises Datacenters Still Important?
While the public cloud has a massive interest, growth, and continued adoption by enterprise customers, the on-premises datacenter is here to stay at least for the foreseeable future. There are simply certainly use cases or regulatory reasons that certain enterprise customers will be unable to adopt the cloud natively. Some customers may have the need to have workloads close to an on-premises server due to “data gravity” reasons. Other customers have legal or other requirements that will not allow housing certain business-critical data in the public cloud.
Traditionally, there has been a problem of overbuying and overprovisioning on-premises hardware so that there is always a percentage left over for potential growth and headroom for other reasons. When you are talking about a scale of hundreds or even more on-premises servers, this can lead to a massive wasting of resources that simply go unused and that may sit idle. It is clear that customers who still use on-premises datacenters need a better way to manage on-premises resources so there is less waste and more efficient management of resources.
Hyper-Converged and Software-defined Solutions
What about hyper-converged solutions that have gained a tremendous amount of momentum in the past few years and various other software-defined solutions?
Hyper-converged solutions are powerful offerings for the enterprise, however, they don’t solve all the needs of today’s businesses. One such limitation of hyper-converged solutions is the scalability of the solution. Most hyper-converged solutions don’t scale beyond a certain number of nodes. Some businesses require much greater scale and flexibility in the number of nodes configured for serving out production workloads.
Additionally, as mentioned, hyper-converged and other popular software-defined solutions that drive hyper-converged infrastructure are generally provisioned for a specific type of workload that is never changes or repurposed. While hyper-converged systems serve certain use cases very well, they do not provide the flexibility in deployment, scale, and other advantages of composable infrastructure used in the datacenter.
What is Composable Infrastructure?
When looking at more traditional approaches to infrastructure and even the more advanced software-defined solutions that drive hyper-converged platforms, there are differences to note. Composable infrastructure follows the design of public cloud delivery of resources in that resources are configured as pools of aggregate resources that can be provisioned for various purposes and workloads. Those resources may also be reclaimed as needed or when no longer in use. The composable infrastructure means resources are “checked out” so to speak and then returned when finished using them. Like the public cloud, resources are first requested and then provisioned based on what is needed for optimum performance.
Public cloud has changed the way that organizations view infrastructure in that it can now be viewed as a service, rather than a tangible “object” in a datacenter on-premises. This provides tremendous flexibility in how those resources can be consumed. The same is true with composable infrastructure. It is a service model approach to infrastructure on-premises – back to the request, provision, reclaim process.
Additionally, when thinking about the now more “normal” or traditional ways of configuring infrastructure on-premises such as using hyper-converged systems, typically these deal with only virtualized resources. The beauty of composable infrastructure is that it encompasses anything from physical, virtual, or containers. The service delivery model allows all of these resources to be added to pools of available infrastructure and then assigned out as needed for delivering production workloads.
How is this management of composable infrastructure accomplished?
You have to have a management layer of software that brokers requests for resources and allows carving out the resources needed from the pools that are available for consumption. This management software layer interacts with the underlying physical/virtual hardware/software by means of a RESTful API. Using API calls, resources can be provisioned near real-time and allocated according to the demands of the workload.
This API management of the infrastructure, including physical, virtual, and containerized applications provides a much more powerful way to manage the pools of resources available. This leads to greater agility, flexibility, efficiency, and the ability to quickly provision needed resources.
Today’s enterprise customers are utilizing public cloud environments for a large portion of their infrastructure needs. However, there are still many use cases where on-premises datacenters are still needed or even required. Following the model that public cloud vendors have pioneered with allowing the consumption of Infrastructure-as-a-Service, composable infrastructure allows applying the same resource pooling methodology to on-premises resources, whether they are physical, virtual, or containerized applications. These can then be efficiently allocated as needed to production workloads and reclaimed at a later time when needed. Additionally, composable infrastructure allows a new mindset of being able to reprovision the underlying resources for different purposes or use cases when needed. Composable infrastructure helps to capture the powerful public cloud service delivery methodology for on-premises environments looking to efficiently control access to pools of resources. By utilizing composable infrastructure approaches to provisioning resources, businesses will operate more efficiently, cost-effectively, and in a service delivery method with its many benefits, all with on-premises resources.Like what you read? Rate us