In part one of this post, we defined SDDCs and discussed how they differ from rigid traditional enterprise data centers as well as their ability to enable a private cloud environment. Today, we will discuss the economic side of SDDCs and how they are portrayed from the perspective of a public cloud provider.
Public cloud SDDCs hold the same principles as their private cloud counterparts. The different data center subsystems (i.e., compute, storage and network) are tightly integrated and run on commodity hardware with central management systems. SDDC technologies have become the de-facto standard in public cloud provider data centers and have been built as “infrastructure as code” from the start due to their sheer size and scalability, with hundreds of thousands of physical commodity blocks. Additionally, heavy CAPEX investments are required, which only web-scale mammoths such as Amazon, Microsoft, and Google, the public cloud market leaders, can afford.
Gigantic Data Centers
In 2014, James Hamilton (a “distinguished engineer” at AWS) described the new capacity that was added by AWS every day as, “enough new capacity to support all of Amazon.com’s global infrastructure.” He wanted to give people an idea of the enormous physical capacity that is added on a daily basis to support the demand for AWS. This impressively fast rate of expansion is not exclusive to Amazon, and can almost definitely be assumed for the other large public cloud providers, too.
The large amount of capital investment involved forces cloud providers to design and operate their data centers in the most efficient way possible. Generally, all cloud providers claim to have the most efficient build-up possible, which enables their “economies of scale” and maintains lower prices for customers. In a recent bid to claim the top spot as the enterprise with the most efficient architecture, Facebook (not a classic IaaS provider, but still…) released its open source data center architecture to the public with a very clear message: “look at how we’re doing it right!”. Leaving no detail to the imagination, the social media giant released everything from high-level blueprints, to their network architecture and building design, all the way down to the detailed spec of the custom pizza box server that they created. Other leaders in the field, however, IaaS vendors, in particular, aren’t following suit in terms of revealing their own blueprints. Nevertheless, regardless of whether or not they are revealing details, all data center giants claim to have unparalleled data center architecture that allows for superior efficiency.
The Initial CAPEX Investment
In order to see returns on their huge capital investments, public cloud vendor on-demand price rates have to embody infrastructure operational costs (e.g., electricity, cooling and administration). Users view the situation a bit differently, however, due to the fact that they make seemingly no capital investment. Users don’t need to build up infrastructure, they don’t have any use for a physical footprint, and they don’t have any fixed costs. Everything is seemingly covered by the provider, but there are exceptions. For example, while it’s not entirely a capital expense, users can make long-term commitments by using reserved instances in Amazon or by leveraging their enterprise agreements for Microsoft Azure.
Another exception is the initial investment that’s required to start building up a cloud infrastructure. This investment includes the cost of constructing a cloud operations team, be it a group that manages a cloud project or hired consultants. Migration costs and other costs that are related to the start-up phase of building a public cloud environment are also capital investments that should be taken into consideration.
In terms of OPEX, as mentioned above, users are under the impression that costs are entirely associated with operational expenses, which are the variable costs that are directly related to usage. We must remember, however, that cloud providers have already embodied their capital investments to build up their infrastructures within the pricing model that users write off as OPEX. The financial aspects of the public cloud have become even more complex for public cloud (IaaS) customers, especially following a recent announcement from the Financial Accounting Standards Board (FASB):
[Disclaimer: While I’m not an accountant, after reading this announcement, I felt that this information was critical for cloud financial managers to know, as well as fellow readers.]
“The Financial Accounting Standards Board (FASB) changed a rule in December  that will make it harder to capitalize the cost of cloud setup and implementations expenses, a change that may encourage some enterprises to opt instead for traditional on-premise software… If a cloud computing arrangement does not include a software license, the customer should account for the arrangement as a service contract.”
We realize the fact that public cloud providers are capitalizing on their investments and that users currently don’t need to be considered when discussing cloud CAPEX vs. OPEX. As public cloud vendors mature, their cloud infrastructure systems become increasingly comprehensive and more able to support efficient operations, which is fundamental to their businesses. The other side of the coin is enterprise IT, which holds its own benefits and challenges, as well. I believe that we are heading into the next stage of the cloud market. While we’re enjoying the significant benefits of the cloud, we are still looking to solve key financial matters.