Cockroach Labs Estate economics What idle nodes, provisioned bytes and peak-sized bills cost, worked out from your estate.
  1. Nodes 127 baseline
  2. Avg CPU utilization 15.0% idle hardware
  3. Storage provisioned 1.5 TB 250 GB of data
  4. Est. monthly saving 0%

Step 1 of 8 - Virtualization

You're running an estate, not a cluster

Every cluster below was sized for its own peak, so every one of them idles.

Deployment model
Separate clusters One Private Host Cluster
25 standalone clusters, 127 nodes in totalcluster-0110% CPU5 nodescluster-025% CPU5 nodescluster-0321% CPU5 nodescluster-0417% CPU5 nodescluster-059% CPU5 nodescluster-0614% CPU5 nodescluster-0710% CPU3 nodescluster-0817% CPU5 nodescluster-0914% CPU5 nodescluster-1012% CPU5 nodescluster-1110% CPU5 nodescluster-1218% CPU5 nodescluster-1319% CPU5 nodescluster-1424% CPU5 nodescluster-1512% CPU5 nodescluster-1613% CPU5 nodescluster-1717% CPU5 nodescluster-1824% CPU5 nodescluster-1913% CPU5 nodescluster-2014% CPU5 nodescluster-2112% CPU7 nodescluster-2216% CPU5 nodescluster-2319% CPU5 nodescluster-2415% CPU7 nodescluster-2518% CPU5 nodesEstate before consolidation25 separate clusters of around 5 nodes each, 127 nodes in total, averaging 15.0% CPU utilization.
15%Average CPUAverage CPU CPU utilization15.0% of provisioned CPU is in use.

127 nodes running 152 vCPU of actual work.

The arithmetic, in full
StepWorkingResult
Provisioned today25 clusters totalling 127 nodes (drawn around 5, 3 to 7) x 8 vCPU1016 vCPU
Mean demand1016 vCPU x 15.0% estate average utilization (per-cluster 5.3% to 24.2%, sd 4.5%)152 vCPU
Aggregate peak152.4 vCPU x (1 + (2.23 - 1) / sqrt(25)), a pooled peak-to-mean of 1.246190 vCPU
Consolidated, sized once for the peakmax(3, ceil(189.8 vCPU / (8 vCPU x 37.5% target))) = max(3, ceil(63.28)); the mean sits well under that peak64 nodes, averaging 29.8%
Auto-scaled through the daythe pool follows the pooled curve in whole nodes, 30 minutes behind it, holding its size for 2 hours before shrinking53 nodes on average, at 35.9% CPU
Utilization across the day30.3% to 39.1% against a 37.5% setpoint: the pool runs hot while it's a bucket behind a rise, and cool while the cooldown is still holding capacity after a peak30.3%-39.1%

Assumptions

  • 8 vCPU per node. Standalone clusters are drawn around 5 nodes, snapped to an odd number for quorum and held between 3 and 15; in this estate they run 3 to 7, a 0.7-node standard deviation, for 127 nodes in total.
  • Each standalone cluster's CPU is a draw from a normal distribution centred on 15.0% and held between 0% and 100%; in this estate they run 5.3% to 24.2%, a 4.5-point standard deviation. The published band is 10-15% today against a 30-45% auto-scaled target, and this Private Host Cluster is set to 37.5%.
  • The Private Host Cluster floors at 3 nodes and must reach 64 at peak.
  • The auto-scaler sizes for what it saw 30 minutes ago and waits 2 hours before shrinking, so the pool averages 35.9% rather than sitting on its 37.5% setpoint.
  • Workloads are independent; consolidation moves where work runs, not how much there is.
What a virtual cluster gives an app team

Its own connection string, databases, schemas, users, roles and backups, and no visibility into the Private Host Cluster or any other virtual cluster.