Like the strangely cruel-to-animals idiom, there are several paths one can embark upon to build a private cloud. What most do not realize is that the journey doesn’t end there. In fact, once the implementation is complete, there’s a whole other (arguably more arduous) task of operating and maintaining that cloud. There are many up-front decisions needed for the on-going operations, management, growth and troubleshooting. Plus, there’s a confusing array tools, with mixed levels of utility, to choose from. Each of these will have a direct impact on the overall cost and length of the journey.
Now when I say operations, it is not just about configuring hardware, monitoring dashboards, and dealing with failures. There’s a myriad of activities that an IT team needs to undertake as they take ownership of an internal cloud. These include:
- Navigating the compatibility lists from various hardware vendors to select the compute, storage and networking elements.
- Determining which software stack to build the cloud infrastructure from.
- Rolling out the initial deployment by integrating the disparate technology parts.
- Adding monitoring and logging to the base environment.
- Training the team on the cloud software stack and infrastructure pieces.
- Day-to-day management of the resource silos.
- Predicting future demand and associated costs for capacity planning.
- Dealing with the onslaught of patches and upgrades from the various vendors comprising the solution.
It’s no wonder that most recently published iteration of the RightScale State of the Cloud Report cites the lack of resources/expertise to manage the cloud as the survey respondents’ top-of-mind concern.
Traditional Private Cloud Economics
Based on conversations I’ve had with enterprise cloud operators, the breakdown of the total cost of ownership of a typical implementation seems to converge on the following split: 15 percent hardware, 40 percent software and 45 percent operations. One’s mileage, of course, will vary – and I would love to hear about your experience and what the distribution is for your environment – but I think a majority would agree that a large portion of the cost is attributed to the operational aspects.
A Self-Operating Cloud: The Next Frontier in IT
The private cloud is a critical ingredient for most data center strategies because, simply put, some workloads are just better hosted on premises. There may be regulatory compliance, data governance and security concerns that preclude the use of the public cloud. Having control over data locality can also have a dramatic impact on application performance and consistency. At large-scale, high-utilization use cases may be more expensive to host on the public cloud once you consider the overall cost in terms of storage, compute and bandwidth over time. Going back to the RightScale report, nearly 80 percent of those surveyed stated that they are incorporating private clouds into their environment. I would argue that, with all these advantages, the adoption of private clouds should be even higher. What’s keeping organizations on the sidelines is the daunting task of on-going operations.
Public cloud vendors have solved the operations challenge by applying intelligent software to enable their IT teams to manage and operate their infrastructure with relative ease. This allows them scale the number of servers per admin to a very large number (usually in the thousands). This is not achievable in most private cloud deployments; those figures are at least an order of magnitude smaller. This makes the economics look much different for enterprise cloud deployments.
The key questions one must consider are: how can we design a cloud that delivers all the benefits of a private cloud, but with a higher-level of operational efficiency? How much productivity would we gain if our infrastructure was able to leverage intelligent software, such as the public cloud, to essentially self-operate and take the day-to-day operational headaches off our plate?
The bottom line: the cloud is a starting point, not the destination. Don’t embark on the journey without deeply considering all the forks in the road and factoring in all the uncertainties at each step.
About the Author
Ajay Gulati is co-founder and CEO of ZeroStack. He leads ZeroStack’s innovative team and corporate strategy. Ajay was a senior architect and R&D lead at VMware where he designed flagship products, including Storage I/O control, Storage DRS and DRS. He has been recognized as a prolific inventor by his peers and has more than a dozen patents to his name. Ajay has a Ph.D. from Rice University and a B.S. from IIT Kharagpur.