One thing is clear to me. Managing IT is similar to playing a game of Jenga, which challenges players to build a stack of blocks then remove those blocks without being the one to knock the whole stack over. Whenever an IT manager wants to make a move or a change to any of the technology stacks in the IT infrastructure, it’s literally like pulling out a block without knocking the entire stack over. For this reason, many companies are taking advantage of cloud computing to automate and simplify their IT management.
However, channel partners with nascent cloud or hybrid IT practices are often unclear about when to use a private cloud, a public cloud, or a combination of the two. For customers, the fundamental questions are when is it best to buy public cloud services? When it is prudent to build your own private cloud? Or does a hybrid cloud make more sense?
The key decision criteria for partners in helping customers to adopt a build (private cloud) vs. a buy (public cloud) approach includes looking at value add, cost, time to deployment, risk profile and experience. After a thorough analysis of these criteria, then partners can engage customers around a conversation of business outcomes and key use cases. Here are some things to consider before having that conversation with your customer.
Private vs. Public Cloud – Utilization Is Key
Typically private cloud solutions tend to work best when the service is core to customer differentiation and driving a competitive advantage. Private clouds also work best if the service is mission-critical, delivers an operational advantage in cost/risk/scale or speed, and if the user experience is directed to the customer.
On the other hand, public cloud solutions work best when the usage is small scale and for temporary or peak projects such as testing, disaster recovery and to manage spikes in business. It’s also the better choice if you are managing a global deployment, and the user experience is for employees. Initially public cloud solutions can seem like a less expensive investment, which is why some IT managers want to move all their workloads over to that platform. But be careful. There are hidden costs associated with it, including storage and I/O, which could add as much as 25% to the cost of a compute service.
Perhaps the simplest way of determining which solution will function best for each workload is to take a look at utilization:
- We are seeing customers owning private cloud infrastructure when utilization is consistently greater than 20 percent. In this scenario, customers will realize the cloud benefits of self-service, flexibility, elasticity, agility, rapid provisioning, greater security, and availability.
- When utilization is consistently less than 20 percent of the time, customers tend to lease public cloud infrastructure so they can see outsourcing benefits of reduced CapEx and a rapid increase or decrease (as the case may be) in capacity.
So a standard rule of thumb for customers assessing cloud is if you are using cloud more than 20 percent of the time, use a private cloud. If you are using cloud less than 20 percent of the time, use a public cloud.
In most cases, the decision comes down to workloads and utilization, and not private vs. public cloud. The ideal scenario is to provision your clouds to accommodate both sustainable and variable workloads. In doing so, there MAY be an incremental 10-15% incremental OPEX spend but a 35-50 percent savings can be realized simply by implementing a hybrid cloud.
Clearly organizations must take on a new approach to evaluate, procure, and deploy IT assets and indeed build new business models enabled through the ubiquity of clouds.
As IT organizations increasingly cope with the demands of rapidly changing business requirements and the need to innovate, they will repeatedly face service-by-service “build versus buy” decisions when sourcing their IT services. And I believe strongly that a hybrid IT approach will most certainly be the primary model for most customers.
Channel partners can serve a valuable role as a trusted advisor for customers who are evaluating how to tackle the cloud opportunity in this new dynamic world of many clouds. Let me know what you think in the comments section below.
Great blog Bob ! Can you elaborate a bit on how you calculate that 20% utilization ?
Thanks Rob. To answer your question, there is a price breakeven point based on the utilization of the public cloud resources (pay per use) and beyond that point it makes sense to use dedicated private cloud resources. For most public cloud customers based on AWS usage pricing that threshold is 20% utilization (based on if the compute resources are used more than 20% of the time, measured over a month).
Based on the book, Cloudnomics, The Business Value of Cloud Computing, written by Joe Weinman, if there is a short enough period of peak demand for cloud services/applications, it makes sense to slice at least that peak demand period out of the total solution and use public on-demand pay-per-use resources to serve it rather than use only dedicated private cloud resources. On the other hand, if there is a long enough duration of non-zero demand, you may as well use dedicated private cloud resources to serve that baseline.