mirror of https://github.com/knative/docs.git
Add load balancing docs (#2559)
* [WIP] Add load balancing docs * feedback from vagababov * fix link * review feedback and removed README * minor tweaks and improvements * minor tweaks, formatting * fixing links * removed note * fix link, default values * review updates * review updates
This commit is contained in:
Ashleigh Brennan
GitHub
parent
647851a051
commit
64d675682c
|
|
@ -1,69 +0,0 @@
|
|||
---
|
||||
title: "Configuring target burst capacity"
|
||||
linkTitle: "Configuring target burst capacity"
|
||||
weight: 50
|
||||
type: "docs"
|
||||
---
|
||||
|
||||
_Target burst capacity_ is a [global and per-revision](./autoscaling-concepts.md) integer setting that determines the size of traffic burst a Knative application can handle without buffering.
|
||||
|
||||
If a traffic burst is too large for the application to handle, the _Activator_ service will be placed in the request path to protect the revision and optimize request load balancing.
|
||||
|
||||
The Activator service is responsible for receiving and buffering requests for inactive revisions, or for revisions where a traffic burst is larger than the limits of what can be handled without buffering for that revision.
|
||||
|
||||
Target burst capacity can be configured using a combination of the following parameters:
|
||||
|
||||
* Setting the targeted concurrency limits for the revision. For more information, see the documentation on [concurrency](./concurrency.md).
|
||||
* Setting the target utilization parameters. For more information, see the documentation on [target utilization](./concurrency.md#target-utilization).
|
||||
* Setting the target burst capacity.
|
||||
|
||||
## Setting the target burst capacity
|
||||
|
||||
* **Global key:** `target-burst-capacity`
|
||||
* **Per-revision annotation key:** `autoscaling.knative.dev/targetBurstCapacity`
|
||||
* **Possible values:** float (`0` means the Activator is only in path when scaled to 0, `-1` means the Activator is always in path)
|
||||
* **Default:** `200`
|
||||
|
||||
**Note:** If the activator is in the routing path, it will fully load all replicas up to `containerConcurrency`. It currently applies target utilization only on revision level.
|
||||
|
||||
**Example:**
|
||||
{{< tabs name="targetBurstCapacity" default="Per Revision" >}}
|
||||
{{% tab name="Per Revision" %}}
|
||||
```yaml
|
||||
apiVersion: serving.knative.dev/v1
|
||||
kind: Service
|
||||
metadata:
|
||||
annotations:
|
||||
name: s3
|
||||
namespace: default
|
||||
spec:
|
||||
template:
|
||||
metadata:
|
||||
annotations:
|
||||
autoscaling.knative.dev/targetBurstCapacity: "70"
|
||||
```
|
||||
{{< /tab >}}
|
||||
{{% tab name="Global (ConfigMap)" %}}
|
||||
```yaml
|
||||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: config-autoscaler
|
||||
namespace: knative-serving
|
||||
data:
|
||||
target-burst-capacity: "70"
|
||||
```
|
||||
{{< /tab >}}
|
||||
{{% tab name="Global (Operator)" %}}
|
||||
```yaml
|
||||
apiVersion: operator.knative.dev/v1alpha1
|
||||
kind: KnativeServing
|
||||
metadata:
|
||||
name: knative-serving
|
||||
spec:
|
||||
config:
|
||||
autoscaler:
|
||||
target-burst-capacity: "70"
|
||||
```
|
||||
{{< /tab >}}
|
||||
{{< /tabs >}}
|
||||
|
|
@ -0,0 +1,33 @@
|
|||
---
|
||||
title: "Load balancing"
|
||||
weight: 30
|
||||
type: "docs"
|
||||
---
|
||||
|
||||
You can turn on Knative load balancing, by placing the _Activator service_ in the request path to act as a load balancer.
|
||||
|
||||
**NOTE:** To do this, you must first ensure that individual pod addressability is enabled.
|
||||
|
||||
## Activator pod selection
|
||||
|
||||
Activator pods are scaled horizontally, so there may be multiple Activators in a deployment. In general, the system will perform best if the number of revision pods is larger than the number of Activator pods, and those numbers divide equally.
|
||||
<!--TODO(#2472): Add better documentation about what the activator is; explain the components of load balancing; maybe add a diagram-->
|
||||
|
||||
Knative assigns a subset of Activators for each revision, depending on the revision size. More revision pods will mean a greater number of Activators for that revision.
|
||||
|
||||
The Activator load balancing algorithm works as follows:
|
||||
|
||||
- If concurrency is unlimited, the request is sent to the better of two random choices.
|
||||
- If concurrency is set to a value less or equal than 3, the Activator will send the request to the first pod that has capacity. Otherwise, requests will be balanced in a round robin fashion, with respect to container concurrency.
|
||||
|
||||
For more information, see the documentation on [concurrency](../../serving/autoscaling/concurrency).
|
||||
|
||||
## Configuring target burst capacity
|
||||
|
||||
Target burst capacity is mainly responsible for determining whether the Activator is in the request path outside of scale-from-zero scenarios.
|
||||
|
||||
Target burst capacity can be configured using a combination of the following parameters:
|
||||
|
||||
- Setting the targeted concurrency limits for the revision. See [concurrency](../../serving/autoscaling/concurrency).
|
||||
- Setting the target utilization parameters. See [target utilization](../../serving/autoscaling/concurrency#target-utilization).
|
||||
- Setting the target burst capacity. You can configure target burst capacity using the `autoscaling.knative.dev/targetBurstCapacity` annotation key in the `config-autoscaler` ConfigMap. See [Setting the target burst capacity](./target-burst-capacity#setting-the-target-burst-capacity).
|
||||
|
|
@ -0,0 +1,76 @@
|
|||
---
|
||||
title: "Configuring target burst capacity"
|
||||
linkTitle: "Configuring target burst capacity"
|
||||
weight: 50
|
||||
type: "docs"
|
||||
aliases:
|
||||
- /docs/serving/autoscaling/target-burst-capacity
|
||||
---
|
||||
|
||||
_Target burst capacity_ is a [global and per-revision](../../serving/autoscaling/autoscaling-concepts.md) integer setting that determines the size of traffic burst a Knative application can handle without buffering.
|
||||
If a traffic burst is too large for the application to handle, the _Activator_ service will be placed in the request path to protect the revision and optimize request load balancing.
|
||||
|
||||
The Activator service is responsible for receiving and buffering requests for inactive revisions, or for revisions where a traffic burst is larger than the limits of what can be handled without buffering for that revision. It can also quickly spin up additional pods for capacity, and throttle how quickly requests are sent to pods.
|
||||
|
||||
Target burst capacity can be configured using a combination of the following parameters:
|
||||
|
||||
- Setting the targeted concurrency limits for the revision. See [concurrency](../../serving/autoscaling/concurrency).
|
||||
- Setting the target utilization parameters. See [target utilization](../../serving/autoscaling/concurrency#target-utilization).
|
||||
- Setting the target burst capacity. You can configure target burst capacity using the `autoscaling.knative.dev/targetBurstCapacity` annotation key in the `config-autoscaler` ConfigMap. See [Setting the target burst capacity](#setting-the-target-burst-capacity).
|
||||
|
||||
## Setting the target burst capacity
|
||||
|
||||
- **Global key:** `target-burst-capacity`
|
||||
- **Per-revision annotation key:** `autoscaling.knative.dev/targetBurstCapacity`
|
||||
- **Possible values:** float (`0` means the Activator is only in path when scaled to 0, `-1` means the Activator is always in path)
|
||||
- **Default:** `200`
|
||||
|
||||
**Example:**
|
||||
{{< tabs name="targetBurstCapacity" default="Per Revision" >}}
|
||||
{{% tab name="Per Revision" %}}
|
||||
```yaml
|
||||
apiVersion: serving.knative.dev/v1
|
||||
kind: Service
|
||||
metadata:
|
||||
annotations:
|
||||
name: <service_name>
|
||||
namespace: default
|
||||
spec:
|
||||
template:
|
||||
metadata:
|
||||
annotations:
|
||||
autoscaling.knative.dev/targetBurstCapacity: "200"
|
||||
```
|
||||
{{< /tab >}}
|
||||
{{% tab name="Global (ConfigMap)" %}}
|
||||
```yaml
|
||||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: config-autoscaler
|
||||
namespace: knative-serving
|
||||
data:
|
||||
target-burst-capacity: "200"
|
||||
```
|
||||
{{< /tab >}}
|
||||
{{% tab name="Global (Operator)" %}}
|
||||
```yaml
|
||||
apiVersion: operator.knative.dev/v1alpha1
|
||||
kind: KnativeServing
|
||||
metadata:
|
||||
name: knative-serving
|
||||
spec:
|
||||
config:
|
||||
autoscaler:
|
||||
target-burst-capacity: "200"
|
||||
```
|
||||
{{< /tab >}}
|
||||
{{< /tabs >}}
|
||||
|
||||
- If `autoscaling.knative.dev/targetBurstCapacity` is set to `0`, the Activator is only added to the request path during scale from zero scenarios, and ingress load balancing will be applied.
|
||||
|
||||
**NOTE:** Ingress gateway load balancing requires additional configuration. For more information about load balancing using an ingress gateway, see the [Serving API](../../reference/api/serving-api) documentation.
|
||||
|
||||
- If `autoscaling.knative.dev/targetBurstCapacity` is set to `-1`, the Activator is always in the request path, regardless of the revision size.
|
||||
|
||||
- If `autoscaling.knative.dev/targetBurstCapacity` is set to another integer, the Activator may be in the path, depending on the revision scale and load.
|
||||
Loading…
Reference in New Issue