Converged and hyper-converged infrastructure is a hot topic at conferences and on tech blogs. These terms are simple: they refer to infrastructure solutions that combine elements (compute, storage and network) to create single solutions that avoid single points failure (especially storage and networking) and simplify the provisioning and support of infrastructure. We’ll be briefly discussing the industry’s move to hyper-converged solutions, and what this means for businesses.
The problem with both hyper-converged and converged infrastructure is that definitions can vary across industries. Potential buyers may find it difficult to see beyond the hype in an industry that is complex.
Converged infrastructure vs. traditional infrastructure
Convergence is a way to simplify the data center. A converged solution includes components like servers, storage, network, etc. This allows businesses to purchase a single piece of a kit that contains the core components necessary to provide cloud infrastructure from one supplier.
This could be a benefit as it requires less effort to build and integrate a system than traditional setups, where companies would have to purchase everything they needed separately, connect it together, and then build it from scratch. VCE, a joint venture between Cisco, VMware, and EMC, began offering converged solutions to help reduce risk, time to market and complexity in building cloud solutions.
For those who want to implement converged or hyper-converged solutions, it is important to review the skills matrix of the team. It is possible for infrastructure changes to cause unexpected disruptions and/or costs.
Converged to hyperconverged infrastructure
The core components of the traditional and converged systems didn’t change during the transition. They were simply packaged together. However, hyper-converged solutions provide a single tier for computing, storage, network, and networking. It is modular and integrates a hypervisor, software-defined storage, and software-defined network. This reduces the number of components that must be managed in the data center.
It is a further simplifying of the data center, which may lead to less physical setup and possibly shorter lead times. However, every company has its own interpretation of hyper-converged, making each solution unique, it is difficult to categorize.
A hyper-converged appliance has a single tier. This means that each server has its own local storage. However, this storage can also be shared among hyper-converged clusters. This pool of storage, also known as a virtual SAN or hyper-converged nodes can be split up and used in smaller volumes as needed.
The solution is simple. You simply add a new block of infrastructure to your existing one. However, this could be a problem.
It’s an all-in one solution. Scalability can be a problem as you will need additional compute. To increase the compute, you will need an entire block (with additional storage and network bandwidth). This might not be an issue for certain types of projects or systems that scale well. This modular approach could lead to businesses being locked into a set of hyper-converged solutions. While these single tiers may work well together, how do you get rid of them once you have, say, 30 tiers? It assumes that your needs will not change. Let’s suppose you have a new, especially resource-intense requirement.