So just when you thought you had gotten your arms around Converged Infrastructure, Converged Platforms and Hyper-converged...
All of a sudden, everyone is starting to talk about Composable Infrastructure (and if you're not hearing about it yet, you soon will and in a big way). Cisco's talking about composable, HPE is releasing a whole new composable architecture called Synergy, and many of us had just figured out how to have a relevant conversation about converged vs. hyper-converged and where they fit. The pace of change is our industry is as rapid as ever, so I thought I'd try to offer a bit of clarity with a high level overview of composable infrastructure and why it matters.
Unfortunately, no one can be told what composable infrastructure is, you have to see it for yourself - wait, that was the Matrix. Actually we can explain composable infrastructure but we need to start with converged and hyper-converged.
So, I suspect most of us are familiar with converged infrastructure. Basically, converged infrastructre referrs to an integrated stack of compute, storage and networking with a unified management layer. Examples include the NetApp/Cisco FlexPod, VCE Vblock, HPE ConvergedSystems, and many others. They can be factory integrated by the supplier (i.e. Vblock and HPE) or certified reference architectures that the customer, VAR or distributor (read Avnet) must integrate before deployment (i.e. FlexPod). More than just a packaging excersize, converged systems accelerate and simplify the design, procurement, deployment and management of data center infrastructure. We are talking about big systems: blade servers, fabric switches and storage arrays. They are designed to provide the power and scalability to run any workload, whether it's virtualized or physical.
Then about 2 years ago we saw the emergence of hyper-converged systems. Hyper-converged systems are single-node systems that can be clustered together to create pools of compute and storage. They are software defined to combine the processors and onboard disk within each node into clusters of compute and storage. The are designed for speed and simplicity, deploying in minutes and easily scaling up (to a point) by simply adding another node to the cluster. Being hypervisor dependent, they are limited to workloads that can be virtualized.
Now, what is composable infrastructure? Composable infrastructure represents the most significant innovation we've seen in the converged space, certainly since the emergence of hyper-converged two years ago, and possibly the biggest innovation to date - period. You could say that composable infrastructure represents a data center architecture with the speed and simplicity of hyper-converged but flexible enough to support any workload (virtualized, physical, containerized). It is a unified architecture consisting of disaggregated pools of compute, storage and network fabric, all controlled by a single management layer. All of the resources are software defined so regardless of where the switches are located or which chassis contain the CPUs or disks, you have the ability to compose and recompose these data center resources as needed. Fluid resource pools that can be provisioned independently and re-allocated at will to support any workload. You will hear this referred to as “infrastructure as code.” Imagine being able to provision your data center infrastructure as if it was racks of processors, racks of storage, racks of network fabric (ethernet, Fiber Channel, iSCSI, FCoE) that can be flexibly managed and consumed as pure capacity based on workload requirements, regardless of whether those workloads are virtualized or bare-metal.
What's the benefit and why now? Composable infrastructure promises to dramatically increase data center efficiency and agility. IT organizations are under increasing pressure to enable business growth as IT is increasingly being viewed as a business driver rather than just a cost center. They are being tasked with becoming much more agile in providing on-demand capacity and support for new ideas, projects and applications. At the same time, IT must maintain the traditional enterprise applications as it always has, ensuring performance, availability and security of the core applications that the business runs on. Composable infrastructure allows IT to meet both challenges by creating a single data center architecture that can support the requirements of traditional enterprise applications, while providing the speed and flexibility to support the demands of new projects and applications in an agile, on-demand manner.
So you can expect to see a lot more buzz around composable infrastructure in 2016, and we're just getting started. Some of the key infrastructure players are already working on the next phase of composable innovation (think breaking apart the processor-memory complex to allow fluid pools of memory that can be provisioned independently from the processor).
If you want to learn more information about the changes in converged infrastructure, I'll be presenting on a webinar next week on Thursday, February 25, 2016 9:00 am Pacific Standard Time. Register here or click below.