Choosing the technologies that populate your newly built data center can be a challenge. There are normally 2 different approaches: Go with what you know works, or try something new. Both come with pros and cons.
In this article, we’ll look at the differences inherent to going with the tried and true old school method of dedicated servers, a fully hyper-converged infrastructure and a hybrid cloud model. While there is really no golden rule for determining the best fit your environment, I’m going to point out a few differentiators that should make you consider the merits of all three.
First, let’s explore the traditional model. In the past, the data center was ruled by dedicated hardware. You’d purchase servers and then configure them for a specific use. Maybe you’d build a Microsoft cluster for SQL or Exchange or perhaps you’d use them for a dedicated application. This worked out fine and, depending on your specific needs, it may still serve well. There are, however, issues with this approach in today’s data-defined world.
While this approach, if implemented correctly, can be highly fault tolerant, it doesn’t scale very well and forces you to calculate and consume resources on an application by application basis. COOP and DR can also be a challenge, requiring you to build a very similar environment at your secondary site. Purchasing servers can, however, be a valid solution for small applications that don’t require a large footprint or need to be kept separate from the overall environment. They can also be rather cost-effective in some cases, again with very small applications. They are not such a good idea when applications have thousands of users and require scale and agility to keep up with consumers.
Secondly, let’s explore hyper-converged. At its most simplified a hyper-converged offering is a single device that provides compute, storage, hypervisor and network connectivity. The goal is to provide a plug-and-play piece of hardware, which either natively or through custom automation can be plugged into an environment to scale the overall capabilities of that environment. The goal is not unlike adding RAM to your PC. You simply plug it in, power it up, and the overall system gains benefit. Hyper-converged infrastructure (HCI) is rapidly becoming the new norm across data centers due to its simplicity and easy scale out.
The US Government has mandated that data centers be consolidated and that administrators move to a more virtualized environment. While some consolidation can be achieved by using dedicated servers and storage, HCI provides a much more simplified approach with guaranteed and predictable scale. Any given HCI appliance has a set of specifications provide a pretty good idea of how many virtual machines it can support. Using that data you can extrapolate the direct benefit to your overall environment. For example, let’s say you know you need to increase the performance of an application by a factor of 20% to stay ahead of user growth. You already know the amount of resources present in the current environment and you know the sizes of the HCI devices available to you. This makes it a very simple matter to calculate how many devices you need to purchase and implement and by extension, you can easily calculate how many additional resources this adds to the environment. This is a very simple explanation, but it is the core of the HCI model.
Lastly, let’s examine hybrid cloud, specifically using hyper-converged and a cloud vendor to create different tiers. (There are many ways to utilize a hybrid cloud infrastructure, which will be covered in other articles, but for this series about the basics, we’ll use a simple example.)
Hybrid clouds provide incredible flexibility when it comes to where and how applications are consumed. Moving to an infrastructure that utilizes both hyper-converged and cloud-based architecture could provide you with unmatched agility and resiliency that other methods simply can’t compete with. By consolidating applications into a hyper-converged infrastructure, then layering in the cloud, you gain the ability to be both extremely efficient and agile. You could perhaps, utilize HCI as your DevOps environment, requiring as little as a single rack of equipment to build, test and deploy new applications then deploy them in the cloud. You can also use the cloud as a backup target, allowing you to have offsite, region independent backups and failover capability. The options available to you by moving to the hybrid cloud are incredibility robust and should be considered one of the most versatile options available in the modern data center.
In closing, I would like to reiterate that the options here are not a “one size fits all,” and you should consider your needs very carefully before building out your data center or upgrading to an HCI infrastructure. One final piece of advice: always consider future needs when making any changes to your environment. I’ve heard it said that you should design based on the environment you’ll have 3 years from now. An approach like that can help you really see the total future value in your purchases, and not just the cost.
The experts here at Swish are well versed in helping our customers get the most out of whatever infrastructure they choose, and we are here to help you make the best choice based on your mission.