Hands-On Workshop: Maintaining EKS Reliability Without Inflating COGS
The Engineering Dilemma: Reliability vs. Efficiency
In the current SaaS landscape, the mandate to reduce Cloud COGS often conflicts with the absolute requirement for application reliability. To avoid performance degradation or OOM events, engineering teams frequently default to over-provisioning resource requests. This "safety buffer" results in low cluster utilization and suppressed gross margins, yet manual right-sizing remains too risky and time-intensive to scale.
On April 30th, CloudBolt and AWS will demonstrate a technical path forward. This workshop focuses on the intersection of EKS Auto Mode and StormForge to show how machine learning can automate the trade-off between performance and cost.
Workshop Agenda & Technical Focus:
Decoupling Safety from Waste: How to move from static, padded resource requests to dynamic, ML-informed limits that respond to real-time production telemetry.
Operationalizing EKS Auto Mode: Hands-on implementation of AWS’s automated infrastructure layer to minimize the manual toil of node group management.
Stabilizing Scaling Logic: Solving the common "thrashing" issues where HPA and VPA conflict.
Measurable Margin Impact: A data-backed approach to reducing Kubernetes-related COGS by up to 70% while improving—rather than risking—system uptime.
Attendee Deliverables:
Direct Lab Experience: Provision and optimize a Kubernetes environment using automated rightsizing tools in a live sandbox.
Hands-On Guidance: Access to AWS and CloudBolt Solutions Architects for a 1:1 look at how a rightsizing practice can address your specific cluster scaling challenges
Full Logistics: Catered breakfast and lunch, and all necessary lab access credentials.
Target Audience:
This session is intended for DevOps Engineers, SREs, and Platform Architects who own the balance between cloud-native performance and infrastructure budget.
Logistics:
Date: Thursday, April 30, 2026
Time: 9:00 AM – 2:30 PM (Lunch provided)
Location: AWS Gen AI Lounge, 525 Market St, Ste 200, San Francisco
Prerequisites: Laptop with kubectl and AWS CLI installed.
Seating is strictly limited to ensure a high-quality, hands-on environment.