$3M+ Annual Cloud Savings in 4 Weeks

Trellis Connect / Trellis Technologies

Trellis Connect was experiencing rapidly rising cloud costs without a clear link to product growth or customer value. Epsilon ran a focused, four-week sustainable cost reduction engagement, addressing root causes across architecture, Kubernetes efficiency, and operational habits. The result was immediate, durable savings with no performance regressions and stronger long-term cost controls.

Sustainable Cost Reduction

Cloud Cost Optimization

Kubernetes Optimization

FinOps Engineering

Infrastructure Optimization

The Problem

As Trellis scaled, cloud spend increased faster than traffic and revenue. Costs were driven by over-provisioned compute, inefficient Kubernetes utilization, orphaned resources, and autoscaling configured for fear rather than data. Leadership needed immediate, meaningful savings without risking reliability or slowing delivery.

Constraints

Production systems could not experience downtime
Cost reduction needed to show results within weeks, not quarters
No performance or reliability regressions were acceptable
Savings needed to persist long-term, not rebound after optimization

Goals

Reduce cloud spend aggressively and safely
Identify and eliminate structural sources of waste
Improve Kubernetes efficiency and scaling behavior
Establish guardrails to prevent cost creep from returning

The Solution

Epsilon executed a four-week, engineering-led cost reduction program focused on architecture, runtime behavior, and operational habits. Rather than relying on surface-level FinOps tactics, the engagement addressed the systems and defaults that caused waste in the first place.

Service-level and workload-level cost attribution
Kubernetes utilization analysis (requested vs actual CPU and memory)
Identification of idle, orphaned, and zombie resources

Right-sized over-allocated CPU and memory across workloads
Adjusted node sizes and instance families to match usage patterns
Deleted unattached volumes, snapshots, load balancers, and IPs

Improved CPU and memory bin packing to increase pod density
Reduced node count while maintaining performance headroom
Tuned autoscaling to eliminate thrashing and overreaction

Established sane defaults for resource requests and limits
Introduced automation to prevent wasteful configurations
Mapped cost ownership to services and teams

The Timeline

4 Weeks; Focused cost reduction sprint with analysis, execution, and guardrails

Week 1

Cost & Architecture Analysis

Built service-level cost model Identified primary sources of waste

Week 2

Right-Sizing & Cleanup

Reduced over-provisioned compute Removed orphaned and idle resources

Week 3

Kubernetes Optimization

Improved bin packing and autoscaling Reduced node counts safely

Week 4

Guardrails & Durability

Established ownership and standards Ensured savings would persist long-term

The Outcome

Within four weeks, Trellis Connect achieved dramatic, measurable cost savings while improving reliability and operational clarity. Savings were realized immediately and continued month over month due to structural improvements and guardrails.

Annualized cloud cost savings

Sustained savings driven by structural changes, not temporary discounts

$3.2M+

Monthly cloud spend reduction

within 4 weeks

~47%

Compute footprint reduction

Across Kubernetes clusters and underlying infrastructure

~38%

Idle / underutilized resources eliminated

~60%

Customer-visible downtime

0

Join the Team

Looking for your next gig?

Get Unlimited PTO, 401(k), Health & Dental Insurance, and many more perks when you join our team.

View Open Positions

Have questions? Our community is here to help. Learn more about our community

Employees

and Counting

We Are Proud to be an
International Employer