💸 ClusterCost Beta
Get Started
← Back to Tools

Pod Packing Simulator

See how your pods stack up. Are you wasting space due to fragmentation?

Add Pods

Stats

Total Capacity 4.0 vCPU
Used 0.0 vCPU
Utilization 0%
Node (m5.xlarge - 4 vCPU)
⚠️

Node Full!

Triggering Cluster Autoscaler...

How to Use This Simulator

  1. Click "Add Pod" buttons: Add small (0.5 vCPU), medium (1.0 vCPU), or large (2.0 vCPU) pods to the node.
  2. Watch the packing: Pods will stack into the available space, just like Kubernetes schedules them.
  3. Hit the limit: Try to add a pod that doesn't fit. You'll see the "Node Full" message—this is when Kubernetes triggers the Cluster Autoscaler.
  4. Observe fragmentation: Notice how leftover space can't always be used efficiently.

What is Pod Packing in Kubernetes?

Pod packing (also called "bin packing") is the challenge of fitting pods of different resource sizes (CPU, RAM) onto a fixed number of nodes. The Kubernetes scheduler tries to pack pods efficiently, but fragmentation is inevitable.

✅ Good Packing

Pods fit perfectly. You use 90%+ of your node capacity. No wasted resources.

Example: 4× 1.0 vCPU pods on a 4 vCPU node = 100% utilization

❌ Poor Packing

Pods leave gaps. You have "space" but can't fit more pods. Wasted money.

Example: 1× 3.0 vCPU pod on a 4 vCPU node = 75% utilization, 1 vCPU wasted

Why Fragmentation Costs Money

Imagine you have a 4 vCPU node. You schedule a 3 vCPU pod. You now have 1 vCPU remaining.

If your next pod needs 2 vCPU, it cannot fit—even though you have "space" on the node. Kubernetes must spin up a new node, and that 1 vCPU on the first node remains idle forever.

💸 The Cost Impact

If you run 100 nodes at 70% utilization due to fragmentation, you're wasting 30 nodes worth of capacity. For m5.xlarge ($140/mo), that's $4,200/month in wasted spend.

How to Reduce Fragmentation

1. Standardize Pod Sizes

Use common denominators for resource requests (e.g., 0.5, 1.0, 2.0 vCPU). Avoid odd sizes like 0.7 or 1.3 vCPU.

2. Use Multiple Node Sizes

Mix small (t3.medium) and large (m5.2xlarge) nodes so oddly-sized pods can find a home.

3. Enable Bin Packing Scheduler

Use the MostRequestedPriority scheduler strategy to pack tightly before spreading across nodes.

4. Monitor Utilization

Use tools like ClusterCost to identify nodes running below 70% utilization and consolidate workloads.

📚 Learn More About Cluster Efficiency