Configuring stateless Apigee hybrid components for production

For production environments, Apigee hybrid stateless components like MART, Runtime, and Synchronizer require specific resource configurations to ensure stability and performance. This topic provides recommended minimum configurations for these stateless Apigee hybrid components in a production installation. Stateless components run in the apigee-runtime node pool, as described in Configuring dedicated node pools.

These recommendations are based on Apigee's default provisioning for paid organizations and should be considered minimum requirements for a production deployment. You may need to adjust these values based on your specific traffic patterns and load testing results.

Production resource recommendations

The following table lists the recommended resource requests and limits for Apigee hybrid components in a production environment.

Component CPU Request Memory Request CPU Limit Memory Limit
MART 500m 2Gi 2000m 5Gi
Watcher 1 4Gi 4 16Gi
Runtime 2 6Gi 4 12Gi
Synchronizer 1 3Gi 2000m 5Gi
Metrics Aggregator 1 10Gi 4 32Gi
Metrics App 500m 10Gi 2 32Gi
Metrics Proxy 500m 1Gi 2 32Gi
Metrics Exporters
(App/Proxy)		 500m 3Gi 1 10Gi
Logger 100m 200Mi 200m 200Mi
Connect Agent 500m 1Gi 500m 512Mi
Mint Task Scheduler
(If using Monetization)		 100m 100Mi 2000m 4Gi

Example overrides

Add the following configurations to your overrides.yaml file to configure your Apigee hybrid installation for production. 

# Production resource requests for MART
mart:
  resources:
    requests:
      cpu: 500m
      memory: 2Gi
    limits:
      cpu: 2000m
      memory: 5Gi

# Production resource requests for Watcher
watcher:
  resources:
    requests:
      cpu: 1
      memory: 4Gi
    limits:
      cpu: 4
      memory: 16Gi

# Production resource requests for environment-scoped components
envs:
  - name: your-env-name
    runtime:
      resources:
        requests:
          cpu: 2
          memory: 6Gi
        limits:
          memory: 12Gi
    synchronizer:
      resources:
        requests:
          cpu: 1
          memory: 3Gi
        limits:
          cpu: 2000m
          memory: 5Gi
  - name: your-other-env-name
    runtime:
      resources:
        requests:
          cpu: 2
          memory: 6Gi
        limits:
          memory: 12Gi
    synchronizer:
      resources:
        requests:
          cpu: 1
          memory: 3Gi
        limits:
          cpu: 2000m
          memory: 5Gi

# Production resource requests for Metrics
metrics:
  aggregator:
    resources:
      requests:
        cpu: 1
        memory: 10Gi
      limits:
        cpu: 4
        memory: 32Gi
  app:
    resources:
      requests:
        cpu: 500m
        memory: 10Gi
      limits:
        cpu: 2
        memory: 32Gi
  proxy:
    resources:
      requests:
        cpu: 500m
        memory: 1Gi
      limits:
        cpu: 2
        memory: 32Gi
  appStackdriverExporter:
    resources:
      requests:
        cpu: 500m
        memory: 3Gi
      limits:
        cpu: 1
        memory: 10Gi
  proxyStackdriverExporter:
    resources:
      requests:
        cpu: 500m
        memory: 3Gi
      limits:
        cpu: 1
        memory: 10Gi

# Production resource requests for Logger
logger:
  resources:
    requests:
      cpu: 100m
      memory: 200Mi
    limits:
      cpu: 200m
      memory: 200Mi

# Production resource requests for Connect Agent
connectAgent:
  resources:
    requests:
      cpu: 500m
        memory: 1Gi
    limits:
      cpu: 500m
      memory: 512Mi

# Production resource requests for Mint Task Scheduler (if using Monetization)
mintTaskScheduler:
  resources:
    requests:
      cpu: 100m
      memory: 100Mi
    limits:
      cpu: 2000m
      memory: 4Gi

Applying the configuration

After updating your overrides.yaml file, apply the changes to your cluster using helm upgrade for each component.

For Helm, apply the changes in order:

Update apigee-operator:

helm upgrade operator apigee-operator/ \
  --namespace APIGEE_NAMESPACE \
  --atomic \
  -f OVERRIDES_FILE.yaml

Apply changes for Logger and Metrics components:

helm upgrade telemetry apigee-telemetry/ \
  --namespace APIGEE_NAMESPACE \
  --atomic \
  -f OVERRIDES_FILE.yaml

Apply changes for MART, Watcher, Connect Agent, and Mint Task Scheduler components:

helm upgrade ORG_NAME apigee-org/ \
  --namespace APIGEE_NAMESPACE \
  --atomic \
  -f OVERRIDES_FILE.yaml

Apply changes for Runtime and Synchronizer components for each environment in your installation:

helm upgrade ENV_RELEASE_NAME apigee-env/ \
  --namespace APIGEE_NAMESPACE \
  --atomic \
  --set env=ENV_NAME \
  -f OVERRIDES_FILE.yaml

See also