Engineering-Led Optimization: How To Proactively Manage Cloud Costs Without Slowing Down Innovation

Historically, cloud cost optimization has been the responsibility of finance teams.

Engineers have bought what they claim to need to build top-notch software; finance teams have gotten (and paid) the bill; neither has had an in-depth understanding of what the other is doing. Without engineering expertise, all finance teams can do to manage cloud bills is take the Costco approach: Buy in bulk, see some savings, and hope it’s good enough.

This has been the status quo in cloud cost management since the late 2000s, when the first third-party cost management platforms came out. As any number of cloud-driven organizations scrambling for efficiency will tell you, the status quo has not done the trick.

Cloud spending continues to mount at an exponential rate. SaaS and cloud-native companies are attempting to pivot from growth-at-all-costs to efficient growth, and have found first-generation cost management platforms unable to facilitate the pivot.

The reason is simple: ​​Every engineering decision is a buying decision. A new cloud resource is a new cloud cost, and engineers are the ones who spin up new cloud resources. For businesses to achieve cloud efficiency, they need to empower their engineers to make smart buying decisions in the cloud.

CloudZero is the first and only platform to make this possible. With our latest feature release — “automated insights,” a suite of platform-driven savings thresholds — we have completed the holistic workflow we call “engineering-led optimization” (or ELO, with apologies to Jeff Lynne).

ELO is how we decentralize cost management to the engineers buying cloud infrastructure, and it’s why customers repeatedly choose us over legacy cost management platforms. The result is straightforward: Engineers make smarter, more efficient buying decisions, achieving maximum value for their cloud spend.

But the underlying process is nuanced. Here, we explain how ELO works — from the first moment of billing ingestion all the way through engineers’ day-to-day work.

How Engineering-Led Optimization Works

1. Ingesting and normalizing 100% of your cloud billing data

Incumbent cost management platforms were built to accommodate a prior generation of cloud consumption — one where AWS predominated, Azure and GCP had some share, Kubernetes had yet to be invented, and specialized services like Snowflake, Databricks, and MongoDB were seedlings.

As cloud consumption has grown complicated, incumbent platforms have struggled to keep up. Most have limitations around what provider data they can present in the platform, and practically none can seamlessly integrate all spend data in a single pane of glass.

If your engineers see anything less than a fully comprehensive, fully explorable account of their cloud spending, there will be cloud spending that falls outside their ability to manage it. So, the first step in ELO is getting 100% of your cloud operational billing data into the platform and normalizing it in a common data model.

The normalization stage of this is extremely important. Different providers’ billing data shows up in different data formats, which is why the vast majority of cost management platforms present different providers’ costs in different views. Normalizing all spend data in a common data model allows us to present it in a single pane of glass, and makes it easy for your engineers to explore and optimize it.

2. Allocate 100% of your spend by business-relevant Dimensions

Complete cloud cost allocation is the foundation of any successful FinOps organization. In order to manage your spend intelligently, you have to be able to quickly determine what’s driving it — and whether it’s necessary.

Traditional cost management has had no meaningful solution for allocation challenges. The result: Tagging gaps, shared resources, multi-tenant architecture, and containerized spend have gated the vast majority of companies’ allocation at 75% or below.

CloudZero CostFormation® is the industry’s only code-driven cost allocation engine. It allows us to allocate 100% of your spend in hours — regardless of the provider, regardless of whether it’s tagged/taggable, regardless of whether it’s shared, multi-tenant, containerized, or any other common allocation challenge — however you want to see it.

It’s what allows us to create custom allocation frameworks for all our customers, depending on how they want to visualize their spend. Cost per customer, cost per product, cost per feature, cost per microservice, and more become possible — and automated — with CostFormation.

3. Attribute 100% of your cloud spend to the engineers/teams responsible for it

If you want your engineers to manage their own cloud spend autonomously, they have to know what they’re responsible for. CloudZero CostFormation also handles this task, attributing 100% of your spend to the correct engineers/engineering teams.

First-generation cost management tools struggle here for the same reasons they struggle with allocation: The best teams in the world now use many shared infrastructure tools, like containers, message queues, shared databases, etc. With no ability to apportion spend to the relevant teams, engineers can’t easily understand the cost impact of their design decisions.

CloudZero’s ability to split up shared spend accurately eliminates the need for a centralized team to monitor shared resources and equips engineers to manage their costs.

4. Make it easy for engineers to get the data they need, when they need it

Engineers are busy. And because first-generation tools were built primarily for finance teams, they did not prioritize an engineering-centric workflow, making it extremely difficult for engineers to get the data they need to manage their cloud costs.

The fewer steps it takes to get the data they need to manage their cloud costs, the more likely engineers will make cost management a regular part of their workflows. CloudZero does this several ways:

CloudZero Explorer: Easy ad-hoc spend exploration

CloudZero Explorer is many engineers’ home base within CloudZero. It’s where they can see 100% of their cloud costs, and filter/drill into any area to understand what’s driving cost trends. It allows engineers to get to the root of a problem within minutes without having to first become a platform power user (customers compare the intuitiveness of CloudZero Explorer favorably to less intuitive features in other platforms).

CloudZero has also recently added role-based access controls (RBAC) which allow engineering leaders to craft specialized views for individual engineers/teams. This makes it easier for engineers to access views that have immediate relevance to their work.

Anomalies: AI-powered spend alerts

CloudZero Anomalies sends alerts directly to the relevant engineers during abnormal spend events. By this stage, we’ve allocated and attributed 100% of your spend, so only the engineers responsible for the affected resources see the alerts, minimizing the “alert fatigue” engineers get when they see alerts not relevant to them.

CloudZero Anomalies compares spend data from the past three days against spend data from the prior 60 days, giving you an extremely granular view of what’s normal/abnormal within the spend Dimensions you’ve established. Best of all, our system handles this whole process — no manual tuning necessary.

Anomalies is one more way we get engineers in the habit of thinking critically about their cloud spending and acting fast when it spikes. Quick wins equals a positive feedback loop, making it more rewarding for engineers to actively manage their cloud spending.

5. Combine human- and platform-driven insights to log and track cloud savings

First-generation platforms center on finance-driven optimizations, offering no meaningful way for engineers to track and log their own optimizations. For engineers to repeatedly engage in proactive cost management, they need this feedback loop — understanding where they can save, how to do it quickly, and how much they’ve helped the company.

CloudZero Insights is the section of our platform where engineers can collect all savings opportunities, collaborate over how to prioritize and remediate each, and track both potential savings and cost avoidance they’ve realized through the platform.

Collect

CloudZero Insights include savings opportunities found by:

Collaborate

Each insight includes meaningful metadata, including estimated cost impact, effort level, and status. Customers use this information to assist in ROI discussions, establish tracking norms across dispersed teams, and facilitate technical discussions between centralized FinOps teams, infrastructure teams, and engineering teams.

Track

CloudZero has a pair of Savings Trackers that automatically quantify the two sides of the savings workflow: before and after savings are realized. Before, insights are added to the “Potential Savings” tracker; after, savings are collected in the “Cost Avoidance” tracker, automatically showing you exactly how much you’ve saved through CloudZero.

Achieve Cloud Efficiency Without Slowing Down Innovation

To sum up the ELO workflow:

First-generation platforms, built for an entirely different era of cloud evolution, simply cannot power this workflow. A new cloud age requires a new cloud platform, one built with cloud buyers — engineers — in mind and at heart.

With accurate, real-time, easily accessible views into the data they’re responsible for; with well-tuned alerts that make it easy to find quick wins; with a comprehensive view of all savings opportunities (and how to realize them); engineers can take total control of their cloud spending — and keep innovating at a rapid pace.

No platform has data as rich and well-organized as CloudZero. No platform combines accountability with intuitiveness as well as CloudZero. No platform equips engineers with the tools they need to manage their own cloud costs as comprehensively as CloudZero. Make ELO your default method of achieving cloud efficiency.

Ready to dig in? Schedule a CloudZero demo today.

Author: Dustin Lowman

Dustin Lowman, a FinOps certified practitioner, takes no small pleasure in applying his English major’s sensibility to the SaaS world. Dustin also writes, performs, and records original music, and has never met a cardigan he didn’t like.