I have had the opportunity to talk with customers over the past few weeks about their plans for converged and hyper-converged infrastructures within their data centers. We usually end up taking a step back and talking about methodologies. Are they comfortable with the idea of software defined yet? Or are they going to stick with a traditional type of infrastructure…
Many consider software-defined data centers to be the next step in the evolution of virtualization and cloud computing as it provides a solution to support both legacy enterprise applications and new cloud computing services. It was sure the talk of the town at the latest Gartner Data Center conference. Hyper-converged infrastructure is a small but rapidly growing segment of the overall CI market, so many customers are looking to do some experimentation in lab environments or try it out on applications that they feel best fit this type of architecture.
Let’s do a quick review of traditional and hyper-converged infrastructures.
Traditional converged infrastructure is an information technology system that packages multiple components into a single optimized IT solution. It typically brings together blade-system servers, enterprise storage arrays, storage area networks, IP networking, virtualization, and management software into a single product. Vblock® Systems are an example of traditional converged infrastructure.
Hyper-converged infrastructure is a software-defined architecture with integrated compute, software defined s
torage, networking, and virtualization. It enables compute, storage, and networking functions to be decoupled from the underlying infrastructure and run on a common set of physical resources that are based on industry-standard x86 components. Using hyper-converged infrastructures, customers can start with a small deployment, and then flexibly scale out to support dynamic workloads and evolving business needs. The VxRack System 1000 rack scale system and VxRail would fit here.
The last point is key when you’re talking about cloud native applications, DevOps environments, or virtual desktops deployments. A hyper-converged rack scale system is the perfect fit for proof of concepts in the core data center because it is capable of growing to support hundreds, even thousands of nodes. The elastic architecture expands easily for an environment like this because compute nodes can be added, removed, or re-allocated as needed without stopping IO. This makes capacity planning and data migrations much easier while also allowing performance to scale linearly as the system grows. This ability to massively scale is what separates the VxRack 1000 from small scale hyper-converged appliances that are out in the market today. Not that the small appliances can’t scale, but it’s how easy they scale and how far they can scale.
Another critical difference is networking. With appliances, that’s usually left to the customer to figure out- both at initial roll-out and as you scale. The VxRack 1000 is a fully engineered system that comes complete with networking- no hassles, no worries.
So let’s dig in and see how the VxRack 1000 can scale-out. This is by far the cool factor that other solutions cannot touch. The ability to add compute or storage as applications and workloads expand or get added to the system. If you need more storage, simply add more storage nodes. If compute is needed, add compute nodes. And this is done independently of each other all the way up to 1000+ nodes. The picture below shows how nodes are added and how it relates to approximate virtual machines. Start small and grow-> 6,000+ vms!
In summary, the VxRack 1000 was another first in the industry from VCE with hyper-converged rack scale systems. Hyper-converged rack scale includes fully integrated networking, standard x86 compute, software defined storage with either EMC ScaleIO or VMware EVO SDDC, and unified provisioning and management. It’s a perfect fit for general enterprise workloads, massive, unpredictable scaling or if you are looking to accelerate the development and delivery of new web services using cloud native technologies.
Stay tuned for future posts. We will look at each individual VxRack System 1000 in detail. Until then, check out the VxRack webpage for more information. We also published a new whitepaper on VxRack System 1000 FLEX for VDI with VMware Horizon View. Contents include components, architecture, benefits, and other details for a deployment of 1,000 linked-clone virtual desktops.