Hyper-convergence has become an increasingly popular way to simplify IT, but it still needs a careful decision process to ensure that what you choose can fulfill the needs of your company.
Hyper-converged technology is not typically viewed as an architecture for a single use case. Many organizations see hyper-convergence technology as a new, general-purpose architecture that supports most or all of the workloads operating in each organization.
But not all enterprises and workloads are identical.
For any new type of architecture available to purchase, you need to spend quality time with your team to investigate the planned system in depth. The system needs to support your organization's current and future workloads, align with your plans (if any) for the cloud and have a price that works for your budget.
When considering hyper-convergence technology, you will face a series of decision points.
Price is often a determining factor in any technology acquisition, even if organizations adamantly deny it. When it comes to hyper-converged infrastructure (HCI), if you're not open to rethinking how you look at the price tag of the IT services you deliver, you may make the wrong selection.
With some hyper-convergence technology products, you may find the direct acquisition cost is greater than simply buying separate servers and storage. Savings can come in other areas, however.
With architecture that fundamentally reshapes how you deliver data center services, considering the acquisition price of hyper-convergence technology alone is not a viable path. You really need to consider your total cost of ownership in order to determine how much the system will really cost. To start, you may be able to reduce the number of people responsible for managing storage, because hyper-convergence effectively eliminates the stand-alone storage layer. As a series of integrated appliances, an IT generalist can manage HCI systems, instead of more expensive specialists. That makes staffing an important consideration in the hyper-convergence buying process.
Do you want to buy a fully integrated, appliance-based infrastructure system from a company such as Nutanix, Dell EMC, Hewlett Packard Enterprise, Cisco, Scale Computing, Pivot3, Datrium or Cloudistics? Or would you prefer to buy just the hyper-convergence software and then build your own hardware configuration? Some of the aforementioned appliance vendors can be talked into selling just software, but there are also companies that do just that, including Maxta, Stratoscale, DataCore and Starwind Software. You can also install VMware vSAN on your server brand of choice or buy in on prepackaged Ready Nodes from the leading server vendors.
It may seem that the only difference between these two approaches would be that one includes hardware and one doesn't. There's a bit more to it, though. In an appliance-based hyper-converged infrastructure, you get what you get. The vendor has a range of hardware options, and you choose one. There is sometimes flexibility in adjusting certain resources, such as buying nodes with increased RAM. With software, though, you get to decide exactly what resource levels you want in each node. And, if you discover after a few months that you need more RAM, you just add it. With some appliances, you're not allowed to add more resources without violating support agreements.
You may wonder why you'd ever buy an appliance. Well, some customers simply don't want to go through the node configuration process. Hyper-convergence technology appliance vendors generally size their hardware to work with a wide variety of applications and to provide consistent performance for customers. If you do go the software route, some hyper-convergence software vendors have deep relationships with vendors that can give you an appliance-like buying and deployment experience.
As the IT industry in general continues its journey to the hybrid cloud, on-premises infrastructure needs to be able to cope. Although it's possible to operate a hyper-converged system without any native cloud integration, doing so is far less elegant than adopting a product that provides native public cloud support.
Most cloud providers offer their own environment to operate workloads. If you implement an on-premises hyper-converged system that doesn't integrate with cloud, as you seek to shift workloads to the cloud, those workloads first must be made compatible with your chosen cloud provider. Those workloads can integrate seamlessly if you pick hyper-convergence technology that is compatible with a cloud provider.
You also need to consider the networking aspect, though. A simple lift-and-shift process won't cut it if the network connectivity isn't there afterwards, particularly if you're using cloud for a disaster recovery scenario. Make sure that the hyper-converged infrastructure product you choose can do the following:
- support your public cloud plans;
- provide the same operating environment for both on-premises and public cloud workloads; and
- enable network connectivity either natively or through the use of a third-party provider.