C-level executives are increasingly calling on IT to investigate cloud options. In fact, according to Gartner, 90% of organizations are looking into crafting a cloud strategy. There are, in fact, significant economic advantages to be found in the cloud. But despite the potential benefits, you need to look carefully before you leap.
Cloud providers have many things working in their favor. For starters, their ability to procure and operate at scale gives them substantial discounts that can be passed along to customers. Cloud customers also benefit by being able to purchase compute for specific applications on a pay-as-you-go basis, without the need for a long-term commitment. The real savings come from elasticity and not having to lay out substantial capital if you only want to ramp up compute for a short period of time.
Additionally, many cloud providers now offer specialized, off-the-shelf services to accommodate common IT needs. So functions like data analytics, elastic load balancing, databases and more can now be plugged into existing environments and used by application developers and IT ops to reduce costs and significantly accelerate deployments.
So with all this in the cloud’s favor, why aren’t all enterprises using cloud today?
If it weren’t for legacy applications, most enterprises would have already migrated. Companies like Netflix were born in the cloud, and thus have architected their applications to run there. As such, Netflix can scale on demand, take advantage of micro services, run across multiple regions in a stateless fashion, and operate elastically. In contrast, legacy applications weren’t built to leverage the cloud’s flexibility, scale, or economic model. What’s more, many were crafted by people long gone, and today’s IT teams are afraid to mess with them.
The typical enterprise is dealing with thousands to hundreds-of-thousands of instances and petabytes of storage embedded in their data centers, co-location facilities, and development labs. Some of their applications will require re-platforming to make them a better fit for cloud and that takes time and money. Is it really worth all that effort to make cloud migration an economically viable proposition?
To explore the cost equation, let’s consider a recent analytical study we conducted in order to identify the best future state options for on-premise or cloud compute. After looking across economic models based on 50,000 OS instances, our analytics concluded that without having to re-platform applications, 45% of those instances would run 36% more cost effectively in the cloud. The enterprises we assessed spend on average $33M per year operating their compute and storage nodes on premise.
So, moving those 45% of instances to the cloud would net $5.6M in annual savings – a 17% reduction, without any need to change applications.
But it’s not always economically advantageous to head for the cloud. When the cost of re-platforming legacy applications exceeds the savings, it may not make short-term sense, but would, we believe, come out positive over the course of five years. Accordingly, we estimate the pace of cloud migration for the enterprise will continue to accelerate incrementally, with between 10% and 20% of enterprise workloads moving to the cloud in 2017.
Still, cloud does not have to be an all-or-nothing proposition, and based on the financial, compliance, and security needs of your applications, instances can operate both on premise and in the cloud. And server refresh is also an option for optimizing costs.
But the bottom line – and the wave of the future – indicates that cloud is a great fit for a large and growing percentage of currently deployed compute and storage infrastructure. The key question then becomes: which instances should you move and how do you find them to bring home the savings?
The answer to these questions can be found by exploring the provisioning, usage, and utilization patterns of your current compute, and then building an unbiased economic model around it. The challenge is churning through hundreds-of-millions of data points to find the best fit for your workloads, while ensuring the results are not tainted by the hand that holds the pencil, so to speak.
To that end, many of today’s leading enterprises, cloud providers and consultants use analytical and algorithmic decision making platforms to assist with the rightsizing and right-costing of enterprise compute. Adopting the right analytical solution can enable your organization to identify the best fit – on premise or cloud, instance size, and pricing option – to accommodate the real-time compute needs of your enterprise workloads.
While analytic and algorithmic analysis solutions will come in various shapes and sizes, to find the right fit there are a few key things to consider. The solution should be one that is cost-effective and does not require new agents to be deployed. Algorithms that have been validated by both hardware and cloud providers will ensure the outcomes are vendor agnostic and always represent your ideal future states. Lastly, the solution should contain a reference library of current on-premise costs, which will allow you to get immediate value and make it easier to identify the best fits for cloud.
By looking at the real-time compute needs of your current workloads and associating it with the economics of ideal future states, you will have the tools to make “To Cloud or Not to Cloud?” An easier question to answer.
This contributed piece has been edited and approved by Network World editors
This article was written by Aaron Rallo, Chief Executive Officer, TSO Logic from NetworkWorld and was legally licensed through the NewsCred publisher network.