The Problem

Polycloud's cloud spend had grown alongside their customer base, which was expected. What was not expected was how difficult it became to understand where that spend was going and which services were responsible for the largest portions of it. Their infrastructure spanned three cloud providers, dozens of regions, and hundreds of services with overlapping dependencies. Getting an accurate picture of what was consuming resources required pulling data from multiple provider consoles, cross-referencing it against internal service maps, and doing a significant amount of manual reconciliation before any meaningful analysis could begin.

The deeper problem was that Polycloud had no reliable map of how their services related to each other across cloud boundaries. A service running on AWS might depend on a data pipeline running on GCP, which in turn consumed resources from an Azure storage layer. When costs spiked, tracing the source of the increase through that chain manually was slow and error-prone. Engineers were spending hours on cost investigations that should have taken minutes, and the conclusions they reached were not always accurate because the dependency picture they were working from was incomplete.

The same gap in dependency visibility that made cost analysis difficult also made change management risky. Deploying a change to a service with cross-cloud dependencies meant accepting uncertainty about downstream effects that the team could not always quantify before the deployment went out.

How Polycloud Used Strand

Polycloud deployed Strand to get a unified dependency map that worked across all three of their cloud environments. For the first time, their engineers had a single view of how services related to each other regardless of which provider they were running on. Cross-cloud dependencies that had previously required manual documentation to track became visible automatically and updated as the system changed.

The impact on cost investigation was immediate. When spend increased in a particular region or service, engineers could trace the dependency chain upstream to identify contributing factors without having to manually cross-reference data from multiple provider consoles. Investigations that had taken hours were completed in a fraction of the time because the dependency context was already assembled.

Polycloud also adopted Strand's change impact analysis as a standard step before any deployment touching services with cross-cloud dependencies. In a multi-cloud environment, the blast radius of a poorly scoped change can extend across provider boundaries in ways that are difficult to anticipate without a complete dependency map. Running impact analysis before deployment gave the team confidence that they understood the full scope of what they were changing before they changed it.

Within six months, Polycloud had used the dependency visibility Strand provided to identify and eliminate redundant resource consumption across their cross-cloud service topology. Several services were found to be pulling data from intermediate layers that were no longer necessary given how the system had evolved, and consolidating those paths reduced cloud spend significantly.

The Results

Cloud spend dropped 60 percent over the six months following Strand deployment. The reduction came from two sources: the elimination of redundant resource consumption that the dependency map made visible for the first time, and the reduction in poorly scoped changes that had previously introduced unnecessary resource usage before being caught and corrected.

Cost investigations became a faster and more reliable process. Engineers working on spend analysis had access to a complete dependency picture from the start rather than having to assemble one manually from multiple provider data sources. The accuracy of their conclusions improved alongside the speed, because they were working from a more complete model of the system.

The pre-deployment impact analysis requirement changed how the team approached changes to services with cross-cloud dependencies. Deployments that previously went out with uncertainty about downstream effects were now reviewed against a live dependency map before approval. The frequency of changes that introduced unintended resource consumption dropped significantly in the quarters following adoption.

Looking Ahead

Polycloud continues to expand across cloud providers as their customer base grows. New services surface in Strand's dependency graph automatically, and the cross-cloud dependency visibility scales with the system without requiring manual maintenance. The engineering team's focus has shifted from reactive cost management to using dependency visibility proactively to make better infrastructure decisions before they affect spend.

"After 180 days in, we had a 60 percent reduction in cloud spend. We knew we had redundant consumption somewhere in the system but we had no way to find it. Strand made the dependency chain visible across all three providers and from there the path to fixing it was clear." Gabriel Santana, Site Reliability Engineer, Polycloud