What is a “private cloud”? That’s a question worth asking these days. Many organizations are pursuing private clouds in order to obtain the agility and cost benefits of public cloud, without incurring the tradeoffs of moving data offsite. Yet despite heavy use of the term, my experiences as an IT strategist at JP Morgan, and now as the head of SimpliVity’s Solutions Architecture team, have taught me both the importance of a true cloud-based model, and the unfortunate prevalence of the same old legacy IT model with a thin veneer of “private cloud” paint. Buyer beware!
Let me start with an analogy. When we think about the cloud, imagine swiping your credit card and walking away with exactly what you need. Delivering that experience comes down to these three pillars:
- Self-service provisioning. Compare options, select a desired service, and turn it on quickly and easily. Aimed at reducing the time to solution delivery, the cloud allows users to choose and deploy resources from a defined menu of options.
- Elasticity to match demand. Pay for what you use, when you use it, with flexible capacity that increases during busy periods, and decreases after demand spikes pass, often using automatic provisioning and load balancing across resources.
- A service-driven operating model to support it. Support, billing, provisioning, and everything else that goes into the service. This allows for increased transparency, and allows consumers to focus on the workloads that drive competitive differentiation, while leaving the basics to a dedicated service delivery team.
Who can say no to that? If you’re not sold yet, think about some of the benefits of this model.
- Reduced cost through improved resource utilization. With a true cloud model, you only pay for what you consume when you consume it, based on a metered usage model. This allows users to deploy more resources during busy times, and give them back when they are no longer needed. This is a dramatic improvement over traditional IT, where long-term capital investments need to be made up front to accommodate expected spikes, leading to unused infrastructure most of the time.
- Faster infrastructure deployment. With flexible, responsive capacity to match demand, cloud offers the potential for much faster deployment times. When you don’t have to start from scratch to build each solution, projects can be greenlighted faster—and with less risk and cost.
- More innovation through low risk trial and error. With cloud, the risk of experimentation can be reduced dramatically. Because you can start small and scale out later, it becomes easier to prototype experimental workloads without high upfront investments. Since you can give back what doesn’t work, you’re not stuck with unneeded equipment if a given experiment doesn’t go well. This has allowed cloud users to take more chances, make more experiments, fail fast, fail cheap, and discover new business opportunities and gain competitive advantage as a result.
Characteristics to look for in private cloud infrastructure
Now that we’ve defined a true cloud model, here are the characteristics you should look for when evaluating private cloud infrastructure. Let’s look at each of them in the context of the three pillars I mentioned above.
- Self-service provisioning
- Orchestration and automation – The infrastructure supporting any private cloud should be fully orchestrated through automation tools like VMware vRealize Automation or Cisco UCS Director, and managed by end users in a given development team, either directly through the tool or via APIs.
- Elasticity to match demand
- Versatility to support any workload – Who knows which of the 10-20 projects the business is investing in will take off? Instead of trying to predict the future, you’ll need a common infrastructure core that can be repurposed for any workload as projects rise and fall at each point time.
- Flexibility to keep pace with changing demand – As demand fluctuates up or down, you’ll need infrastructure that can quickly scale in or out accordingly. Look for fast, simple deployment, and the ability to create policies or move workloads at the VM-level, without worrying about the underlying physical infrastructure.
- Service-driven operating model
- Low upfront investment – If you’re introducing a metered usage model, you don’t want to make a huge investment upfront until you have a sense of what demand will look like. Ideally, you’d want to start with a small upfront investment to keep utilization close to demand until your needs expand.
- Easy to divide up workloads for metered billing – Complicated infrastructure with multiple vendors and lots of fingers in the pie make it difficult to carve up workloads and allocate an accurate price per VM. Look for infrastructure that can streamline chargebacks by making it easy to allocate a percent of utilization per project or workload.
Lastly, it’s critical to consider the cost of delivering this kind of model. If achieving everything outlined above costs 10 times more than the market leading public cloud providers, odds are your private cloud initiative will never get funded. Look for a solution that provides a cost per VM that is close to public cloud. It’s okay if your private cloud is a few dollars more per VM. The important thing is that cost cannot be claimed as an advantage to choosing public cloud. Given all the other benefits that a private cloud delivers, most enterprises will be happy to spend a few dollars more to keep their data on-premises.
So there you have it. That should help see through the façade of any would-be cloud imposters.
In the next installation of this blog series, I’ll move on to talk about how SimpliVity helps its customers deliver true private cloud business models, along with some specific customer examples.
In the meantime, you can learn more about SimpliVity hyperconvergence and private cloud by checking out our new white paper.