Cluster Right-Sizing Recommendation API

Cluster Right-Sizing Recommendation API

GEThttp://<your-kubecost-address>/model/savings

Kubecost's Cluster Right-Sizing Recommendation API can monitor the resource utilization of your clusters and offer cost-effective right-sizing solutions.

Table 1. Path Parameters
Name Required Type Description
window false string Duration of time over which to query. Accepts multiple different formats of time (see this Using the window parameter section for more info).
targetUtilization false float in the range (0, 1) Target CPU/RAM utilization which parallels environment profiles. For reference, Development should equal .80, Production should equal .65, and High Availability should equal .5. Also supports custom values within the range.

minNodeCount

false

int

Minimum node count to be recommended which parallels environment profiles. For reference, Development should equal 1, Production should equal 2, and High Availability should equal 3. Also supports custom values within the range.

allowSharedCore

 false

boolean

Whether you want to allow shared core node types to be included in your recommendation. Accepts true or false.

architecture

 false

string

Accepts x86 or ARM. Currently, ARM is only supported on AWS clusters.
Table 2. Responses
Code Description Example
200 OK 
{
   "code": 200,
   "data": {
   "cluster-one": {
   "recommendations": {
   "multi": {
   "pools": [
   {
   "type": {
   "provider": "GCP",
   "name": "f1-micro",
   "vCPUs": 0.2,
   "RAMGB": 0.6,
   "hourlyPrice": 0.0076,
   "spotHourlyPrice": 0.0035,
   "sharedCore": true,
   "architecture": "x86",
   "region": "us-west1",
   "pricePerRAMByteHr": 3.946013748645783e-12,
   "pricePerCPUCoreHr": 0.031611,
   "spotPricePerRAMByteHr": 8.307397365570068e-13,
   "spotPricePerCPUCoreHr": 0.006655
   },
   "count": 3,
   "totalMonthlyCost": 16.644,
   "totalRAMGB": 1.7999999999999998,
   "totalVCPUs": 0.6000000000000001
   },
   {
   "type": {
   "provider": "GCP",
   "name": "n1-highcpu-2",
   "vCPUs": 2,
   "RAMGB": 1.8,
   "hourlyPrice": 0.049594019999999996,
   "spotHourlyPrice": 0.014915600000000001,
   "sharedCore": false,
   "architecture": "x86",
   "region": "us-west1",
   "pricePerRAMByteHr": 2.7622096240520478e-12,
   "pricePerCPUCoreHr": 0.0221277,
   "spotPricePerRAMByteHr": 8.307397365570068e-13,
   "spotPricePerCPUCoreHr": 0.006655
   },
   "count": 1,
   "totalMonthlyCost": 36.203634599999994,
   "totalRAMGB": 1.8,
   "totalVCPUs": 2
   }
   ],
   "nodeCount": 4,
   "monthlySavings": 20.5239354,
   "totalMonthlyCost": 52.84763459999999,
   "requiredRAMGB": 2.7075592279434204,
   "totalRAMGB": 3.5999999999999996,
   "utilizationRAMGB": 0.752099785539839,
   "requiredVCPUs": 1.9305600746179477,
   "totalVCPUs": 2.6,
   "utilizationVCPUs": 0.7425231056222875
   },
   "single": {
   "pools": [
   {
   "type": {
   "provider": "GCP",
   "name": "n1-standard-1",
   "vCPUs": 1,
   "RAMGB": 3.75,
   "hourlyPrice": 0.033249825,
   "spotHourlyPrice": 0.01,
   "sharedCore": false,
   "architecture": "x86",
   "region": "us-west1",
   "pricePerRAMByteHr": 2.7622096240520478e-12,
   "pricePerCPUCoreHr": 0.0221277,
   "spotPricePerRAMByteHr": 8.307397365570068e-13,
   "spotPricePerCPUCoreHr": 0.006655
   },
   "count": 3,
   "totalMonthlyCost": 72.81711675,
   "totalRAMGB": 11.25,
   "totalVCPUs": 3
   }
   ],
   "nodeCount": 3,
   "monthlySavings": 0.5544532499999946,
   "totalMonthlyCost": 72.81711675,
   "requiredRAMGB": 2.2937504053115845,
   "totalRAMGB": 11.25,
   "utilizationRAMGB": 0.20388892491658528,
   "requiredVCPUs": 1.800551426660732,
   "totalVCPUs": 3,
   "utilizationVCPUs": 0.6001838088869107
   }
   },
   "parameters": {
   "clusterId": "cluster-one",
   "clusterName": "",
   "staticVCPUs": 1.4105254827890852,
   "staticRAMGB": 1.0523239374160767,
   "nonDaemonSetPodCount": 20,
   "daemonSetVCPUs": 0.13000864795721567,
   "daemonSetRAMGB": 0.41380882263183594,
   "daemonSetCount": 5,
   "maxPodVCPUs": 0.26021418473138547,
   "maxPodRAMGB": 0.13831710815429688
   },
   "preferences": {
   "minNodeCount": 3,
   "strategies": null,
   "targetUtilization": 0.8,
   "allowSharedCore": true,
   "architecture": ""
   },
   "currentClusterInfo": {
   "monthlyRate": 73.37156999999999,
   "nodes": [
   {
   "name": "e2-medium",
   "provider": "GCP",
   "architecture": "x86",
   "count": 3,
   "RAMGB": 4,
   "vCPUs": 1
   }
   ],
   "totalCounts": {
   "totalNodeCount": 3,
   "totalRAMGB": 12,
   "totalVCPUs": 3
   }
   }
   }
   }
   }

Examples

Receive right-sizing recommendations taking into account cluster activity for the past two weeks, and ideal CPU and RAM utilization for a high availability environment.

http://<your-kubecost-address>/savings/clusterSizingETL?range=2w&targetUtilization=.5&minNodeCount=3&allowSharedCore=false&architecture=x86

Receive recommendations taking into account AWS cluster activity for the past five days, ideal CPU and RAM utilization for a production environment, and allowing shared cores.

http://<your-kubecost-address>/savings/clusterSizingETL?range=5d&targetUtilization=.65&minNodeCount=2&allowSharedCore=true&architecture=ARM

Adopting cluster right-sizing recommendations

The Cluster Right-Sizing Recommendation API is not able to directly implement its recommendations. To adopt right-sizing recommendations for your cluster(s), view this section of our Cluster Right-Sizing doc.