What makes the public cloud so powerful is the ability to run short experiments involving a large number of compute resources, something that is too expensive and takes too much time to procure in the data center. 

Specifically, there are so-called non-deterministic polynomial-time problems where known algorithms exist that can find good solutions, however mathematically no best solution can be guaranteed. Examples for these types of problems are Sudoku, a number puzzle, protein folding, the traveling salesman problem, and the knapsack problem.

Software engineers can implement known approximation algorithms in the cloud to see if they will produce viable solutions to real-world problems. Once a proof of concept has been found, the cloud solution can be optimized to make it commercially viable. This will result in a competitive advantage for your business and is the main reason why we don't want to reduce the velocity of innovation by building guardrails instead of gatekeepers in the cloud.

Gartner forecasts that the worldwide public cloud services market will grow to $364 billion USD in 2022, while Forrester estimates the public cloud market will grow to $411 billion USD by 2022. The cloud is changing our businesses. Every company is becoming a technology company in addition to their core industry. Think of Nike, Nationwide Insurance, and Walmart.

But the cloud has many challenges. We are going to look at these challenges from a technology, management, and financial perspective. When we look back to the data center there was a lot of waste because workloads were statically provisioned. Engineers set up buffers to accommodate changes in demand, leadership set up buffers on top of these to accommodate growth, and executives set up additional buffers to mitigate risk. All of this waste was unavoidable due to the lack of elasticity in the data center.

However the cloud is a managed service where someone else is applying software patches and carrying a pager, and using the same buffering strategies will make it more expensive compared to the data center. Forbes says executives estimate that at least 30 percent of their cloud spending is wasted. Unfortunately, cloud technology is still in its early stages and there is generally no native functionality where idle workloads are being suspended to avoid incurring cost.

When we look at the cloud from a management perspective we see that even medium-size businesses may have millions of rows in their cloud bill. And cloud vendors maintain millions of stock-keeping units or SKUs across hundreds of services they offer. For example, Amazon Web Services Elastic Compute Cloud has over 4 million SKUs. Traditional spreadsheet methods cannot handle this volume, making it necessary that customers use specialized tools to manage things like security, compliance, deployment, cost, and waste at cloud scale.

When we look at the third-party vendor market for cloud, there are many companies offering tools, sometimes with overlapping functionality. While we see some consolidation through acquisitions, there is currently not one tool that does everything we need. This makes it necessary for companies to use multiple tools to manage different aspects of the cloud.

And when we look at the cloud from a Finance perspective we see that the centralized approval model from the data center shifted to a decentralized model where engineers are making purchasing decisions. Because software engineers typically don't have a background in Finance, we need new methods to hold them accountable for the cost of their workloads.

Financial Operations or FinOps aims to hold engineers responsible for costs similar to DevOps, where engineers became responsible for operationalizing their workloads. FinOps is primarily a culture shift where cloud cost is moved to the forefront of everyone's thinking. And this is what we will focus on in our lecture.

To summarize the challenges of Cloud FinOps: The cloud is a game-changing technology, where companies are becoming technology companies in addition to their core business. Elasticity replaces fixed spend models with variable cloud consumption. Ownership is decentralized as it shifts from a central organization to engineers. And the cloud is inefficient, as few cloud providers offer native tools to turn off workloads when they are not in use.