Upravljate troškovima svojih cloud resursa? Ne? Ne brinite, niste jedini. Jako malo organizacija adekvatno upravlja. Trude se. Ignoriraju. Ne zamaraju se. A troškovi kontinuirano rastu. Zatim se čude i pišu senzacionalističke članke na koje sam se referencirao u ovom tekstu.

Cloud usluge su zamišljene da su skalabilne i dostupne u trenutku kada ih zatrebate. A, naravno, tako se ponašaju i troškovi – skalabilno. I trenutno. Onog trena kada zatražite cloud instancu. Bez obzira da li ju koristite ili ne. Ako je upaljena, naplaćuje se.

FinOps kultura i principi su zamišljeni da pomognu u upravljanju cloud resursima i vezanim troškovima tako da maksimiziraju balans vrijednosti i troška. Kroz tri faze zatvorenog ciklusa – inform, optimize, operate – cilj FinOps metodologije je kontinuirano praćenje korištenja resursa te predlaganje optimizacija (primjerice rightsizing) te drugih aktivnosti (primjerice rezervacije resursa) kako bi maksimalno iskoristili cloud resurse.

Iako sam prošao službenu edukaciju i prošao certifikaciju (malo se hvalim :)), knjiga je obavezno štivo kao dopuna edukaciji. Prepuna je korisnih savjeta, prijedloga a i iskustava iz stvarnih situacija i tvrtki.

Cloud FinOps

Part I – Introducing FinOps

“In the simplest terms, FinOps brings financial accountability to the variable spend model of cloud”

“It’s the way for teams to manage their cloud costs, where everyone takes ownership of their cloud usage supported by central best-practice group”

“With FinOps, each operational team can access the near-real-time data they need to influence their spend and help them make data-driven decisions that result in efficient could cos balanced against the speed/performance and quality/availability of services”

“In our experience, you should provide engineers with feedback on the impacts of their actions as close as possible to the time those actions occur”

“Like DevOps before it, FinOps is a cultural shift, and the earlier it starts, the sooner an organization will benefit”

“When you are focused on speed and innovation, it’s easy for cloud bill to soar”

“A central FinOps enablement team is the drive of a FinOps mindset across the organization evangelizing best practices for incorporating cost consideration into development plans”

“It’s important to note that you can’t facilitate a cultural shift to FinOps just by practitioners or bringing in a contractor. With FinOps, everyone in the organization has a part to play”

“The disconnect comes when these teams [operations, finance] try to communicate with each other, which cloud forces them to do so much more often than was common during the data center days”

“FinOps should help the finance teams to understand the cloud language by abstracting the cloud-specific terminology away for those who understand the dollars and cents, and they should simplify the finance requirements for the engineering teams”

“Understanding this complexity [cloud billing] allows you to do really cool things down the road, like automated anomaly detection when something small changes. We say small because cloud efficiency often suffers a death by a thousand cuts”

“Remember, you’re being charged for the actual time in the period that the thing was on. Not whether it was used, but whether it was on”

“But in the cloud, compute resources should be thought of as ephemeral and replaceable. They are cattle, not pets”

“The most successful FinOps practices decentralize using less (i.e. avoiding costs) and centralize paying less (i.e. reducing rates)”

“Give the engineers/application owners the data and the power to make the right decisions for their business objectives”

“FinOps isn’t about saving money, it’s about making money through data-informed decisions about your cloud technology investments”

“Your adopting FinOps plan should simply outline the steps the business will take to start-or-expand-FinOps and the benefits of getting the practice going”

“Every organization adopting the cloud will need to develop mechanisms to manage cloud costs. The Framework provides an operating model for your FinOps practice by aligning cloud management best practices”

“Your reports and dashboards must present information consistently, with high data quality and with high degree or review and control”

“Making your FinOps reports accessible to all your staff is essential”

“Your [FinOps] job is to ensure each respective team has the context to truly understand cloud consumption data and, when faces with it, make the right assumptions leading to the right actions for the business”

“The reports and dashboards you create are some of the most important tools in building a collaborative FinOps practice”

Part II – Inform Phase

“The cloud is a value creator, but the more you use it, the more cost it will incur. The role of FinOps is to help maximize the value created by that spend”

“Real-time decision making is about getting data – such as cloud spend changes or anomaly alerts – quickly to the people who deploy and manage cloud resources”

“The FinOps team improves the available data via better tooling and modifies the business processes that enable your organization to do FinOps”

“These phases [inform, optimize, operate] aren’t linear – you should plan to cycle through them constantly”

“The inform phase is where you start to understand your costs and the drivers behind them”

“During the inform phase, you get visibility into cloud spend, drill down into granular cost allocation, and create shared accountability”

“Some of the activities in this phase include: Map spending to the business, Create showback (and other reporting), Define budget and forecasts, Define your account strategy, Set tag strategy and compliance, Allocate shared costs equitably, Dynamically calculate custom rates and amortizations, Analyze trending and variance, Create scorecards, Benchmark against industry peers, Identify anomalies”

“The optimize phase identifies measured improvements to your cloud and sets goals for the upcoming operate phase”

“This [optimize] phase includes the following activities: Analyze KPI’s and set goals, Find and report underutilized services, Evaluate centralized commitment-based discount options”

“The operate phase sets up the processes for taking actions to achieve those goals [improvement]”

“This phase [operate] also stresses continuous improvement of processes”

“Here are some of the activities that take place during the operate phase: Deliver spend data to stakeholders, Make cultural changes to align with goals, Rightsize instances and services (or turn them off), Define governance and controls for cloud usage, Continuously improve efficiency and innovation, Automate resource optimization, Integrate recommendations into workflows, Establish policy-driven tag cleanup and storage lifecycle policies”

“Tagging spend metrics to business metrics is a key step in your FinOps journey”

“Managing anomalies starts in the inform phase. If you are not looking at your anomaly reporting, stop reading and go turn it on, then come back and finish this chapter. It’s that important”

“Using scorecards that benchmark team performance is the best way to find out how teams operate”

“If that weren’t enough, a solid tagging strategy allows you to build more accurate forecasting models, simply because you’re using data that’s richer, more accurate, and more granular”

“Showback leaves the bill on the table for analysis; chargeback leaves the same bill, but requires payment”

“Tags allow you to assign business-context details to specific resources and then to analyze that information to answer your company-specific questions”

“The key to winning at FinOps is to et account and tagging strategy established early on and maintain them consistently”

“The most flexibility and success comes from having and approach that encompasses both accounts and tags”

“When you’re leveraging tagging for cos allocation, you might initially focus on tags for the business unit, the product, the owner, and the role. Then you might ensure your development services are deployed into different account than your production services. You use accounts to isolate costs by environment”

“Tagging is vital early step that ensures you’ll have a rich dataset to work with when you get into more advanced stages of your FinOps practice”

“If you’re unsure of where to start with a tagging standard, here are some tags that companies with successful FinOps implement: A cost center/business unit tag, A service workload name, A resource owner, A name tag, An environment tag”

“We recommend using an account hierarchy in place of the most important one or two tags – say, cost center and environment – and than using tags for the remaining three to five allocations that must be tracked”

“To determine how your tagging practices are working inside your organization, you should have reports identifying where tags are incorrectly applied. We recommend you measure tag average by the amount of cloud spend with a tag allocated versus without”

“It is done [forecasting] to understand how future cloud infrastructure and application lifecycle changes may impact budgets and to aid evaluating cloud investment decisions. And it’s one of the hardest things to get right in FinOps”

“Forecasting at finer levels of granularity can enable you to determine where your forecasts are not matching the actuals and allow you to focus on those areas to improve”

“The most advanced practitioners in the State of FinOps reported updating their forecasts weekly, though monthly or quarterly was still the most common cadence for the majority”

“When you introduce FinOps optimizations, the reduction of your operating cost may hide the natural growth of your cloud spend”

“The general consensus in the FinOps community is that cloud team forecasts should be done on a rolling 3-to-12 month basis”

Part III – Optimize Phase

“Goals aren’t a single value -they’re a trend over time. If you set a goal as a single value such as x dollars per month, you have to wait until the end of the month to see how you are tracking”

“Adding a target line allows you to get more context about the graph data points”

“This is what we classify as an expected anomaly. A decision is made ahead of time to exceed the planned budgets to achieve a desired business outcome”

“The goal isn’t to make perfect decisions [for usage optimizations]. It’s to make the best decision possible given the context and data, and then reevaluate often. A well-sized resource won’t stay that way forever, as code and load evolve over time”

“When cloud resources are first deployed, often their capacity requirements aren’t fully known or well understood. That’s true for almost everyone in those early days of deployment”

“The central FinOps team is there to operate the automated analytics into resource usage, provide reports on what resource appear to be underutilized, and programmatically provide alternative, more efficient configurations”

“When looking at the existing workload on a resource (like a server instance), it’s crucial to ensure that you’re not just looking at average and peak values of the metrics. If you use averages or peaks, you can be misled”

“It is more cost-effective to rightsize and discount your workloads, thereby potentially leaving some unused discount capacity, than it is to ignore rightsizing for fear of risking wasted discounts”

“Put simply, serverless can be cheap, but refactoring for serverless is not”

“We suggest starting with the lowest-impact recommendations first, such as cleaning unused volumes and IP addresses. Test out rightsizing recommendations in non-production environments, and get the broader organization used to the variable usage benefits that cloud provides via it’s ability to change resource sizing to match needs”

“Assigning tickets to team members generally makes them feel more accountable for cost wastage”

“However you decide to calculate savings and track them over time, do it consistently, document your assumptions, and have it ready. Because one thing most FinOps practitioners report is that sooner or later the leadership is charge of the FinOps team will stop by to ask how much the FinOps team has saved by optimizing, and they usually want an answer right away”

“Use cases for owned/perpetual licenses or license rights purchased for non cloud to be used in the cloud are extremely complex and are often subject to changing rules and requirements”

“This is [most popular] because commitment based discounts represent the largest percentage discount you can achieve in cloud and often apply to the largest areas of cloud spend in your bill”

“RIs, SPs, CUDs, and Flexible CUDs are the most complex optimizations available to cloud users. There are serious discounts to be gained by operating a successful reservation program”

“Cloud is designed for just-in-time purchasing. You shouldn’t be running infrastructure or purchasing capacity (i.e. buying commitments) in a data center fashion”

“When a commitment doesn’t match the usage you are expecting, it can lead to very large cost increases”

“When you are earlier on in the maturity of your [FinOps] practice, we recommend buying just one commitment. This allows you to confirm your knowledge with low-risk purchase”

Part IV – Operate Phase

“Whereas the optimize phase sets the goals, the operate phase takes action”

“Start by figuring out how items get added to the new process (onboarding), what is expected from the teams (responsibility), how teams are informed about when the process should be followed and the outcomes of following the process (visibility/alerting), and finally begin following the process (action)”

“Teams should understand how their contributions help the organization reach goals. When teams don’t see how their efforts contribute, they often don’t follow processes”

“When it comes to any business tooling, the inevitable debate begins: buy versus build. For FinOps we recommend starting with a third-party tool”

“You shouldn’t try to implement too much optimization all at once. This will inevitably cause issues and have you struggling to make sure the automation you deploy is working effectively”

“The primary rule od MDCO [Metric-Driven Cost Optimization] is: don’t do anything. That is, don’t do anything until you have a metric that measures the impact of your action”

“Remember, a cloud service provider can only observe metrics provided by the VM hypervisor, not operating systems or processes, unless you’re using their managed service offering”

“In the container world, you can tag/label your containers. These tags form the identification that you use to allocate costs”

“Namespaces allow you to group resources within a single cluster. Creating namespaces for teams, or for specific services, allows you to create a coarse-grained cost allocation group”

“The alternative is to run multiple [Kubernetes] clusters, but running too many clusters can be hard to manage and isn’t cost-efficient”

“FinOps is a team sport, and very little gets done without the engineers”

“A simple definition for unit economics metrics is direct revenues or cost, associated with a particular business model, that are specifically expressed on a per-unit basis”

“Don’t be discouraged if you’re unable to get to a top-line revenue metric related to cloud spend. Many organizations cannot, either because of the type of business they are in or because of their maturity”