scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: {{ $gateway.name }}
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: {{ $gateway.cpu.targetAverageUtilization }}
---

  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: {{ $gateway.name }}
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: {{ $gateway.cpu.targetAverageUtilization }}
---

If you have an API that does a comparable amount of computations every time and you have a lot of clients, then the **CPU utilization** will probably *also be stable* (instead of constantly going up and down quickly).

### Examples of Replication Tools and Strategies

				if v.N > 0 {
					di.Metrics.LastMinute[k] = v.asTimedAction()
				}
			}
			for k, v := range info.Metrics.APICalls {
				di.Metrics.APICalls[k] = v
			}
			if info.Total > 0 {
				di.Utilization = float64(info.Used / info.Total * 100)
			}
			disksInfo[index] = di
			return nil
		}, index)
	}

	g.Wait()
	return disksInfo
}

// Get an aggregated storage info across all disks.

new APIs so we can get out of alpha.

 *  New: Support Java 21's virtual threads (‘OpenJDK Project Loom’). We changed OkHttp's internals
    to use `Lock` and `Condition` instead of `synchronized` for best resource utilization.

 *  New: Switch our Internationalized Domain Name (IDN) implementation to [UTS #46 Nontransitional
    Processing][uts46]. With this fix, the `ß` code point no longer maps to `ss`. OkHttp now embeds

	// in the 10 node shard would coordinate between other 9 shards to get the
	// rest of the `99*9` requests.
	//
	// This essentially splits the workload properly and also allows for network
	// utilization to be optimal, instead of blindly throttling the way we are
	// doing below. However the changes that are needed here are a bit involved,
	// further discussion advised. Remove this comment and remove the worker model

   ([#123492](https://github.com/kubernetes/kubernetes/pull/123492), [@carlory](https://github.com/carlory))
- Added metric name along with the utilization information when running `kubectl get hpa`. ([#122804](https://github.com/kubernetes/kubernetes/pull/122804), [@sreeram-venkitesh](https://github.com/sreeram-venkitesh))

!!! note "Remarque"
    Les sections de ce chapitre ne sont **pas nécessairement "avancées"**.

    Et il est possible que pour votre cas d'utilisation, la solution se trouve dans l'un d'entre eux.

## Lisez d'abord le didacticiel

Tout est basé sur la déclaration de type standard de **Python 3.8** (grâce à Pydantic). Pas de nouvelles syntaxes à apprendre. Juste du Python standard et moderne.

Si vous souhaitez un rappel de 2 minutes sur l'utilisation des types en Python (même si vous ne comptez pas utiliser FastAPI), jetez un oeil au tutoriel suivant: [Python Types](python-types.md){.internal-link target=_blank}.

Vous écrivez du python standard avec des annotations de types:

Vous devez vous assurer qu'il est unique pour chaque opération.

```Python hl_lines="6"
{!../../../docs_src/path_operation_advanced_configuration/tutorial001.py!}
```

### Utilisation du nom *path operation function* comme operationId