Sections in this category

Container Request Right-Sizing Recommendation API (V1)

NOTE: This API is deprecated. Please use the V2 API if starting new work.

The container request right-sizing recommendation API provides recommendations for container resource requests based on configurable parameters and estimates the savings from implementing those recommendations on a per-container, per-controller level. Of course, if the cluster-level resources stay static then you will likely not enjoy real savings from applying these recommendations until you reduce your cluster resources. Instead, your idle allocation will increase.

The endpoint is available at

http://<kubecost-address>/model/savings/requestSizing

Parameters

Name Type Description
targetCPUUtilization float in the range (0,1] An amount of headroom to enforce with the new request, based on the calculated (real) usage. If the calculated usage is, for example, 100 mCPU and this parameter is 0.8, the recommended CPU request will be 100 / 0.8 = 125 mCPU. Inputs that fail to parse (see https://pkg.go.dev/strconv#ParseFloat) or are greater than 1 will not error; they will instead default to your savings profile’s default value. If you have not changed the profile, this is 0.65.
targetRAMUtilization float in the range (0,1] Calculated like CPU.
window string Duration of time over which to calculate usage. Supports hours or days before the current time in the following format: 2h or 3d. See the Allocation API documentation for more a more detailed explanation of valid inputs to window. Defaults to 1d.
filterClusters string Comma-separated list of clusters to match; e.g. cluster-one,cluster-two will return results from only those two clusters.
filterNodes string Comma-separated list of nodes to match; e.g. node-one,node-two will return results from only those two nodes.
filterNamespaces string Comma-separated list of namespaces to match; e.g. namespace-one,namespace-two will return results from only those two namespaces.
filterLabels string Comma-separated list of annotations to match; e.g. app:cost-analyzer, app:prometheus will return results with either of those two label key-value-pairs.
filterServices string Comma-separated list of services to match; e.g. frontend-one,frontend-two will return results with either of those two services.
filterControllerKinds string Comma-separated list of controller kinds to match; e.g. deployment,job will return results with only those two controller kinds.
filterControllers string Comma-separated list of controllers to match; e.g. deployment-one,statefulset-two will return results from only those two controllers.
filterPods string Comma-separated list of pods to match; e.g. pod-one,pod-two will return results from only those two pods.
filterAnnotations string Comma-separated list of annotations to match; e.g. name:annotation-one,name:annotation-two will return results with either of those two annotation key-value-pairs.
filterContainers string Comma-separated list of containers to match; e.g. container-one,container-two will return results from only those two containers.

Savings Projection Methodology

The request right-sizing recommendation includes an estimate of the savings that can be realized by applying the request right-sizing recommendations. To calculate this estimation, we use each container’s lifetime and the overall data window (max observed cluster lifetime within window). We assume each container will run on the same node (and therefore have the same resource costs) it ran on historically; calculate the monthly rate for that container with the new, reduced resource requests; and then we scale that monthly rate by container lifetime in window/data window. This will underestimate savings for recently-created controllers (e.g. a Deployment created 3 days ago in a 7-day data window will be assumed to run for 37 of the next month when calculating monthly savings), but avoids some edge cases that vastly overestimate savings.

Savings projection examples

Two Pods, each with their own controller

We have a 1 hour window with 2 pods that look like they each have their own controller. Each pod has 1 container (with the same name).

All CPU costs are $7/core-hour

Pod 1 ran for 15 minutes [t=15min, t=30min], allocated 3 cores, and used an avg and max of 1 core.

Pod 2 ran for 20 minutes [t=45min, t=60min], allocated 3 cores, and used an avg and max of 2 cores.

|   ---      | Pod 1 exists
|         ---| Pod 2 exists
|____________|
  time ->
|            |
0 min        60min

Window = [0min, 60min]

We’ll right-size with a target utilization of 100%:

  • Pod 1 will be right-sized to an allocation of 1 core.
  • Pod 2 will be right-sized to an allocation of 2 cores.

What should the estimated monthly savings of this right-sizing be?

Controller 1 = Pod 1 ran for (1545) of the known duration of the cluster being alive (we don’t know if it was alive from [t=0, t=15]). That’s (45 min / (60 min/hr) / (730 hr/month)) of a month.

Within the query window, the pod could haved saved: 2 cores * (15min / (60 min/hr)) = 0.5 core-hours 0.5 core-hours * $7/core-hour = $3.50

“If that 45 minute window is representative for 30 days (730 hrs) then we scale the savings by 1 / (45 / 60 / 730)”: $3.50 * 1 / (45 / 60 / 730) = $3406.67

For Pod 2 = Controller 2 we can take the same numbers from Pod 1 = Controller 1 and halve the savings because it has half the CPU core savings.

Savings: $3406.67/mo / 2 = $1703.34/mo

Total savings = $3406.67/mo + $1703.34/mo = $5110.01/mo

The above, but the Pods share a controller

We resize the shared container to 2 cores, reducing the savings of pod 1 to be the same as the savings for pod 2, because both pods had the same overall allocation.

Controller 1 = Pod 1 and Pod 2 ran for 4545 minutes of the known duration of the cluster being alive (we don’t know if it was alive from [t=0, t=15]). That’s (45 min / (60 min/hr) / (730 hr/month)) of a month.

Within the query window, Pod 1 could haved saved: 1 cores * (15min / (60 min/hr)) = 0.25 core-hours 0.25 core-hours * $7/core-hour = $1.75

Pod 2 saves the same amount = $1.75

That’s a total savings for the controller of: $1.75 * 2 = $3.50

“If that 45 minute window is representative for 30 days (730 hrs) then we scale the savings by 1 / (45 / 60 / 730)”: Total savings = $3.50 * 1 / (45 / 60 / 730) = $3406.67/mo

API Examples

KUBECOST_ADDRESS=http://localhost:9090

curl -G \
  -d 'targetCPUUtilization=0.8' \
  -d 'targetRAMUtilization=0.8' \
  -d 'window=3d' \
  ${KUBECOST_ADDRESS}/model/savings/requestSizing

Edit this doc on GitHub