Optimization settings and descriptions on StormForge Documentation

Annotations

Mon, 01 Jan 0001 00:00:00 +0000

Goal

live.stormforge.io/cpu.optimization-goal: "Balanced"
live.stormforge.io/memory.optimization-goal: "Balanced"

Schedule

live.stormforge.io/schedule: "@daily"
live.stormforge.io/schedule.timezone: "UTC"
live.stormforge.io/validity-period: ""

live.stormforge.io/auto-deploy: "Disabled"
live.stormforge.io/learning-period: "P7D"
live.stormforge.io/auto-deploy.thresholds.cpu.percent: ""
live.stormforge.io/auto-deploy.thresholds.memory.percent: ""
live.stormforge.io/auto-deploy.thresholds.cpu.unit: ""
live.stormforge.io/auto-deploy.thresholds.memory.unit: ""
live.stormforge.io/apply.method: "PatchImmediateRollout"
live.stormforge.io/apply.max-percent-increase: ""
live.stormforge.io/apply.max-percent-decrease: ""

Reliability

live.stormforge.io/reliability.oom.memory-bump-up.period: "P4D"
live.stormforge.io/reliability.oom.memory-bump-up.percent: "0"
live.stormforge.io/reliability.oom.memory-bump-up.min: "0Mi"
live.stormforge.io/reliability.oom.memory-bump-up.max: "2Gi"
live.stormforge.io/reliability.oom.memory-bump-up.apply-immediately: "Never"

CPU and memory

live.stormforge.io/containers.cpu.optimization-policy: "RequestsRaiseLimitsIfNeeded"
live.stormforge.io/containers.memory.optimization-policy: "RequestsRaiseLimitsIfNeeded"
live.stormforge.io/containers.cpu.limits.limit-request-ratio: "2.0"
live.stormforge.io/containers.memory.limits.limit-request-ratio: "2.0"

live.stormforge.io/containers.cpu.requests.min: "10m"
live.stormforge.io/containers.cpu.requests.max: ""
live.stormforge.io/containers.memory.requests.min: "32Mi"
live.stormforge.io/containers.memory.requests.max: ""

Optimization goal

Mon, 01 Jan 0001 00:00:00 +0000

StormForge always recommends values that minimally satisfy both savings and reliability, but you can set goals to shift the balance in favor of one or the other based on your risk tolerance and optimization goals.

The optimization goal applies to the entire workload.

Optimization goal

Used in determining recommended values based on your goals and risk tolerance.

Annotation	Default value
`live.stormforge.io/cpu.optimization-goal`	`"Balanced"`
`live.stormforge.io/memory.optimization-goal`	`"Balanced"`

Description

The optimization goal for a resource indicates which of two competing criteria to maximize when computing recommended request values: cost savings or workload reliability. StormForge will always produce values that minimally satisfy both criteria, but the goal setting can be used to shift the balance in favor of one or the other.

Schedule

Mon, 01 Jan 0001 00:00:00 +0000

Following the learning period, StormForge generates recommendations at the interval defined by the schedule. An optional recommendation validity period can be configured independently from the recommendation schedule.

For more details and examples showing how these settings are related, see the Schedule, Validity period, Learning period, and Preliminary recommendations sections in the Concepts topic.

Recommendation settings:

Schedule

The schedule on which StormForge generates new recommendations for a workload.

Annotation	Default value
`live.stormforge.io/schedule`	`"@daily"` (equivalent to `P1D`)

Description

Recommendations are generated automatically for workloads on a schedule, defined by a schedule string. The shortest valid recommendation schedule is once an hour.

Apply settings

Mon, 01 Jan 0001 00:00:00 +0000

Define how StormForge deploys generated recommendations.

Apply settings:

Deploy recommendations automatically

These settings define whether StormForge deploys new recommendations automatically.

Auto-deploy

When "true", StormForge automatically applies new recommendations when they are generated on the configured schedule.

Annotation	Default value
`live.stormforge.io/auto-deploy`	`"Disabled"`

Description

When auto-deploy is enabled (see Examples below), StormForge automatically applies new recommended settings when they are generated on the configured schedule.

Reliability response

Mon, 01 Jan 0001 00:00:00 +0000

As changes in memory usage occur, StormForge detects and reports out-of-memory (OOM) events on all workload types as it collects metrics. You can configure StormForge to respond to OOM events by applying a temporary memory bump-up and indicate when the bump-up is to be applied.

Note:

StormForge tracks and reports OOM events on all workload types. However, at this time, temporary memory bump-ups are supported for Deployments and ReplicaSets only. Memory bump-ups are not supported at this time for StatefulSets, DaemonSets, and CronJobs.

CPU and memory

Mon, 01 Jan 0001 00:00:00 +0000

CPU and memory settings are applied at the container level. The configuration used can vary from container to container within a workload.

CPU and memory settings:

For details about how to set these values using annotations, see one of these topics: Configure clusters, Configure namespaces, or Configure workloads.

Container-level settings syntax

When specifying configuration values for CPU and memory settings, you can specify a default value for all containers, one or more values for specific named containers, or a comma-separated list.

HPA

Mon, 01 Jan 0001 00:00:00 +0000

StormForge integrates with Horizontal Pod Autoscalers (HPA) and can:

Generate vertical rightsizing recommendations for HPA-managed workloads
Preserve HPA scaling behavior when adjusting resource requests

How StormForge integrates with HPA

StormForge integrates with the HPA by adjusting HPA target utilization when Pod resource requests change.

Because HPA scaling is based on a percentage of resource requests, changing requests without adjusting target utilization changes how a workload scales.

StormForge balances vertical sizing changes with corresponding adjustments to HPA target utilization so the workload continues to scale at the same overall load level.

Runtime

Mon, 01 Jan 0001 00:00:00 +0000

Java runtime

You can define how StormForge applies JVM max heap size recommendations to JVM containers.

The Java max heap recommendation settings apply to workloads identified as JVM workloads that StormForge is able to collect JVM metrics for. For non-Java workloads, these settings will have no effect.

StormForge by default assumes that Java applications use the -XX:MaxRamPercentage heap size management mechanism, and that changing the container’s memory limit will therefore change the effective max heap size. StormForge will attempt to use this relationship to optimize heap size without changing any configuration for your app.

Pod scheduling

Mon, 01 Jan 0001 00:00:00 +0000

Scheduling optimization policy

Whether or not node affinity is included in recommendation patches depends on the optimization policy selected.

Annotation	Default value
`live.stormforge.io/pod-scheduling.optimization-policy`	`DoNotOptimize`

Description

StormForge recommends an instance category for pod scheduling by default, but that recommendation will not be included in patches unless an optimization policy is configured.

Instance category node affinity recommendations can be added to patches by setting the optimization policy to InstanceCategories.