Hello again, Jon Slusser here to present the first module of the Cloud FinOps and Pricing workshop. In this module, we address the field of Cloud FinOps in cost modeling. As such, we have the following learning objectives : understand how the basic economics of cloud are changing and impacts to organizations in forecasting, budgeting, and cost management, explain the basics and importance of cloud cost modeling, including total cost of ownership, marginal cost and the challenges an organization faces in crafting their cost mops, introduce the emerging field of Cloud FinOps in its practice areas, understand best practices for public Cloud cost management and how visibility, culture, and governance are important levers, identify key personas involved in Cloud economics and their unique concerns. Then let's get started. The changes brought by Cloud are not limited technology, there's been a disruption in the economic aspects as well. As compared to traditional hosting models enjoyed by enterprises over previous decades, the emergence of Cloud has disrupted these models while also delivering fresh opportunities. The traditional models relied on heavy fixed capital investments sustainable stakes taken into the game. These investments take many years to amortize consume with poor decisions leading to unused capacity and potentially large write-downs. Because budgeting cycles tend to have long intervals, lead times are unnecessarily long with low responsiveness to market demand signals. This financial model lends itself well to hierarchical control, providing fertile ground for centralized governance models, which typically went along with limits on developer autonomy and independence. The cloud model has brought forward significant change. With limited upfront investment, the financial hurdles to achieving their presence are minimal and lower risk financially. The pay-as-you-go model results in the ability to stop spending on a dime and divert resources to more promising initiatives immediately. Due to the approachability of the Cloud, the time from idea to instantiation happens in near real-time. Ultimately, the result is democratization of Cloud consumption decisions, time devalue, and developer autonomy. This is, it turns out, can be a double-edged sword if not managed properly. This graph illustrates a key concept of the economic model change brought by public cloud. In the traditional hosting model, the fixed upfront investment is significant and takes relatively high volume to justify. However, with sufficient volume, incremental usage in the private model has a marginal cost approaching zero. The public model is very approachable with minimal upfront fixed spend, the cost tend to increase linearly with increases in consumption volume. The value of the public PAYGO model are obvious, speed and innovation. On the other hand, further volume increases continue to drive costs in a linear fashion, ultimately becoming more expensive than private at some volume crossover point. As we've seen, many organizations, while realizing great value from their public cloud investments, found they were spending more in public cloud than they expected. This is not to say we're resigned to paying higher costs for public cloud, as we will see through this workshop, there are numerous methods and techniques we can use to manage and optimize the cost. This brings us to the topic of cost modeling. Organizations with mature finance and product management practices will have models that measure the cost of their services and applications. This brings many benefits, including supporting conversations like these. What does the application or service cost versus what should it cost? Are we realizing value of the app or service that exceeds its cost? Where should we host the app or service, and are there less costly alternatives? If we EOL the service or moved it, where do we save money? Credibility is key, merely going through the exercise and producing data that isn't broadly believed to have integrity will do more harm than good in supporting good investment decisions. According to the 2019 Bain's survey of enterprise and small and medium businesses, total cost of ownership remains a primary criteria for Cloud migration decisions. This increases the emphasis on having cost models and a cost management plan when operating in the public cloud. Another interesting finding is a dramatic increase in interest in shifting models from CapEx to OpEx to exploit the increased financial flexibility this brings. What challenges do organizations face with the cost model? Poor quality data is often an obstacle as all but the most disciplined organizations suffer with inventory tracking and CMDB process compliance. Often telemetry tooling is inconsistently deployed, if at all, particularly for organizations that are distributed, which leads to incomplete or even the absence of important data such as resource utilization. This may force the use of theoretical rather than actual data, which in turn will affect model credibility. While patterns across companies are often similar, model reuse should be avoided. The here use this model often isn't sufficient for anything but an informed starting point. Because there's not a single correct model, in fact, all models are wrong to some degree, a good enough point must be reached to avoid over-investment in the creation of the model. Also, variations and flexibility in the model are important. Using the right version model to answer the question being asked is critical. Let's use an example, say an HR department is considering the EOL of an on-prem application to save hosting costs, what is reasonably saved on infrastructure this application is EOLD? If you're using a total cost of ownership model, the answer will be overstated. In reality, savings are probably minimal if any, and limited to variable costs alone. This because the fixed costs of private infrastructure are sunk and not recoverable. At best, one can repurpose, recapacitate the other applications, if there are such candidates. A marginal cost model should be used to answer this question instead. Let's extrapolate this example to another question. If I migrate this HR Application to the public Cloud, will it cost us less? The answer is it will probably cost more in aggregate. This is because the cost of hosting a public Cloud represents net you spent, likely well above the marginal cost of keeping the application where it is. Therefore a decision to migrate to public, it should be based on criteria other than cost savings alone. Here's another example from Intel's own enterprise IT organization. This cost modeling exercise has been performed each year for the last decade with an attempt to proactively answer the question of whether the enterprise private Cloud if migrated to public, would save Intel money. The model has been tuned over the years and speaks to an average unit cost of an IS instance hosted in our private Cloud. This is a TCO model which captures all components from facilities to support to licensing. This model does not answer the questions of app modernization approaches as it focuses on a lift and shift scenario. Indeed, the industry has seen that simple lift and shift approaches to public Cloud usually result in net higher spends for reasons that are now intuitive. Failure to capture benefits of application modernization and failure to address cost optimization and process in the public Cloud once there. This isn't an indictment of public Cloud. We are using public Cloud for reasons far beyond saving money. An approach that does not focus on cost optimization and modernized consumption models is likely to disappoint. The challenges of Cloud cost management have become pervasive enough that the emergence of formalized practices have occurred. FinOps.org is one example of a collaborative organization that promote structure, process, and practices of Cloud cost management. Their resources treat the topic with substantially more depth than we deliver here. In addition to collaboration, they offer training and certification for those inclined. Here's a typical pattern we see an organization spin-off practices as they progress in their Cloud journey. In the ad hoc phase, there's early adopt to public Cloud use, often organic and perhaps the aggregate spending on public Cloud is starting to get the attention of leadership. Standardized phase is characterized by higher volume public Cloud usage with initial attempts to implement some form of governance and control. Common examples include formation of a Cloud Center of Excellence, Cloud brokers practices, and a centralized procurement approach. The procurement centralization can be accompanied by formalized master agreements or contracts with the Cloud service providers that provide some form of discounting. The optimized phase is an aspirational state where all personas in the organization are incentivized for economic efficiency in the solutions they impact. Financial metrics are measured, are widely visible and there's cost accountability across the organization. Understanding where our customers are in this journey inform the approaches we can use to assist and help them to make an impact on their cost efficiency in the Cloud. What are some of the best practices that enable organizations to mature in their FinOps practices? In general, one must start with the identification of no more than a handful, relevant metrics to keep the focus on cost efficiency and not dilute with less important measurements. Supporting data to produce these metrics will include sources such as Cloud billing reports, application inventory, and infrastructure utilization. This is sometimes an art of the possible where we do the best we can with the data that is available. This ability, the metric, should be elevated to prominence and broadly disseminated. Accountability and incentives are required across all personas to drive alignment to the purpose. Driving accountability is dependent on a culture of cost optimization which will need to be set at the top. Governance can support this culture through organizations, civic bodies while the Cloud Center of Excellence is one method that may be employed. Billback is a method that is extremely effective in supporting the cost optimization culture and reduces the reliance on strict governance to achieve the desired results. This table illustrates typical personas you'll encounter at accounts and what their interests and limitations are relative to Cloud cost management. Exact personas are going to lead the direction in culture development and reinforcement but are distant to the design and real-time consumption decisions. Architects can support by assuring systems designs include data standards and telemetry tooling to support continuous cost optimization. The largest gap is typically at the DevOps developer site, reliability engineer standpoint as incentives at this persona are typically at odds with cost-efficiency as these folks were rewarded for reliability, functionality, and sufficient capacity. Your finance and procurement folks will be adept at negotiating discounts and contracts but usually you're not influential in the topic of continuous optimization. Some forward leading organizations aren't creating a formal FinOps role who is charged with bridging architecture DevOps and finance and an overall cost optimization approach. Bottom-line, all these personas have a role to play in implementing their culture and practice of Cloud cost management and optimization. As a summary of what we discussed in this module. While Cloud isn't primarily about saving money, Cloud cost management is a critical and training priority for organizations. Construction and proper use of cost models is a critical input to business decisions. Organizations will have different degrees of maturity in the Cloud journey through Cloud cost management. Measurement, visibility, and accountability are essential pillars of Cloud cost optimization, while governance and culture are important levers of change in persistence. We reviewed common personas and organizations that will be evolved in Cloud cost management and what their unique contributions and concerns are. This concludes the module.
