Case study

Flaschenpost keeps Kubernetes clusters 15–25% smaller

How Flaschenpost uses Flexera Ocean to continuously rightsize AKS workloads with zero downtime.

Flaschenpost case study

At a glance

Company: Flaschenpost
Industry: Freight & Logistics Service
Location: EMEA | Münster, Germany
Company size: Enterprise
Primary use case: Kubernetes rightsizing / Cloud Cost Optimization
Products used: Flexera Ocean / Autonomous Workload Optimization (formerly Spot Ocean and Elastigroup) 

Executive summary

Flaschenpost, a fast-growing German quick commerce company, runs its platform on Azure Kubernetes Service (AKS). Developer-owned resource settings created costly over-provisioning, while manually tuning hundreds of applications could not scale. Flaschenpost uses Flexera Ocean, also known as Autonomous Workload Optimization, to manage its node fleet and continuously match workload CPU and memory to actual usage. The company estimates its clusters would otherwise be 15 to 25% larger. In-place rightsizing also reduces manual effort and, on Kubernetes 1.33 and above, applies changes without restarting pods. 

Featured results

  • Cluster efficiency: estimated 15–25% smaller clusters than without rightsizing
  • Service continuity: zero downtime for critical services with in-place rightsizing
  • Operational efficiency: no manual audits across hundreds of applications
  • Resource accuracy: CPU and memory matched continuously to actual workload usage
  • Cost governance: reduced excess capacity and more predictable infrastructure costs

“It's not just about cost savings, it's about giving your team the ability to focus on higher value work rather than constantly managing infrastructure.” 

Denis Pesikov

Denis Pesikov SRE and DevOps Engineering Manager, Flaschenpost

The challenge

Over-provisioning raised Kubernetes costs

Flaschenpost gives developers ownership of deployment settings, including CPU and memory requests, limits, and horizontal scaling. That autonomy helps teams move quickly, but it often produced over-provisioned workloads that became costly at scale. Resource tuning at the workload level required a large manual effort and could not keep pace as the platform grew.

Key challenges included:

  • Developer estimates could become inaccurate or outdated
  • Node and workload capacity both required optimization
  • The platform team faced manual review across hundreds of applications
  • Excess requested capacity increased infrastructure cost 

The solution

Automated rightsizing of Kubernetes workloads with zero downtime

Flaschenpost's over-provisioning problem existed at two levels: the nodes, meaning the machines that provide compute, and the workloads, meaning the applications running on those machines. Flexera Ocean addressed both.

Ocean handled the nodes first. It managed Flaschenpost's node fleet automatically and packed workloads onto nodes efficiently, so the clusters ran well without the team managing compute by hand. That took care of one layer, but the workloads themselves were still overprovisioned, and tuning them one by one was manual work that couldn't keep up as the platform grew.

Flexera Ocean’s in-place rightsizing feature solved that. Once Flaschenpost turned it on, it started matching each workload's CPU and memory to what the workload was using, continuously and in the background. For Denis's team, that changed the day-to-day in a few concrete ways:

  • No more manual audits. Developers stopped constantly tuning their workloads, and the team no longer had to review hundreds of applications by hand.
  • Live updates, zero downtime. On Kubernetes 1.33 and above, rightsizing applies changes without restarting pods, so Flaschenpost's critical services keep running uninterrupted.
  • Fewer errors. Ocean sets resources based on usage at the workload level, instead of developer estimates that can be inaccurate or outdated.

What had once been a repetitive, manual task now ran on its own. Flaschenpost has used Flexera Ocean for at least four years and was among the first EMEA customers to run Ocean for AKS in production.

“In practice, this has dramatically reduced operational overhead for my team. Developers no longer need to constantly tune their workloads, and we don’t need to manually audit hundreds of applications.” 

Denis Pesikov

Denis Pesikov SRE and DevOps Engineering Manager, Flaschenpost

The result

A leaner, more efficient Kubernetes environment with predictable costs

Autonomous Workload Optimization's rightsizing delivered measurable results for Flaschenpost. The company estimates that without it, its clusters would be roughly 15 to 25% larger than they need to be. By rightsizing every workload to its usage, Ocean removes that excess capacity and drives substantial cost savings across Flaschenpost's infrastructure.

The impact reaches well beyond cost. With workload tuning handled automatically, Flaschenpost's team reclaimed time, reduced errors, and gained confidence that workloads would scale efficiently as the business grew. This has helped the company sustain operational excellence and advance its cloud cost efficiency agenda, giving Flaschenpost predictable cost governance that supports continued platform growth without losing control of infrastructure spend. Because Ocean optimizes resources continuously, it also reinforces reliability, allowing the team to maintain high service performance with no manual intervention.

"It's not just about cost savings, it's about giving your team the ability to focus on higher value work rather than constantly managing infrastructure." 

Denis Pesikov

Denis Pesikov SRE and DevOps Engineering Manager, Flaschenpost

BUSINESS IMPACT 

Continuous Kubernetes optimization built for FinOps at scale

For Flaschenpost, in-place rightsizing has become a core part of how it runs Kubernetes, and Denis Pesikov sees the same potential for any organization operating at scale. By keeping workloads continuously optimized, reducing manual effort, and improving resource utilization, the feature aligns directly with the cloud efficiency and FinOps goals that Flaschenpost continues to pursue. As the platform grows, that ongoing optimization is what allows the team to spend less time managing infrastructure and more time on the work that moves the business forward.

“For anyone considering Flexera Ocean, I would strongly recommend enabling the in-place rightsizing feature. From my experience, it’s a solution that pays for itself quickly and becomes an integral part of running a reliable, cost-efficient cloud platform.”

Denis Pesikov

Denis Pesikov SRE and DevOps Engineering Manager, Flaschenpost

Next steps

Keep Kubernetes workloads optimized as your platform grows