> For the complete documentation index, see [llms.txt](https://developer.harness.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://developer.harness.io/harness-platform/3.0/harness-platform-resources/overview/strategy/concurrency.md).

# Concurrency

Limit how many strategy instances run simultaneously when a stage or step uses a matrix, for loop, or while loop strategy.

**Concurrency** controls how many instances run at the same time when a [strategy](/harness-platform/3.0/harness-platform-resources/overview/strategy.md) is applied to a stage or step. Without a concurrency limit, all instances start simultaneously. Setting a limit queues the remaining instances and releases them as running ones complete.

Use concurrency to:

* Avoid overwhelming a downstream service with simultaneous deploys
* Respect infrastructure quotas or rate limits
* Roll out changes progressively across environments or regions

## What you will learn

* What the `max-parallel` setting does and how it limits simultaneous strategy instances
* How to configure a concurrency limit in YAML alongside a matrix or loop strategy
* The behavioral difference between no limit, serial (`max-parallel: 1`), and N-parallel execution

***

## YAML

Set `max-parallel` inside the `strategy` block:

```yaml
stages:
  - name: deploy
    id: deploy
    strategy:
      matrix:
        region: [us-east-1, eu-west-1, ap-southeast-1, ap-northeast-1]
      max-parallel: 2
    steps:
      - name: deploy-region
        id: deploy_region
        run:
          script: echo "Deploying to ${{ matrix.region }}"
```

With `max-parallel: 2`, the four regions deploy in two waves: the first two start immediately, and the remaining two start only after one of the first two finishes.

***

## Behavior

| `max-parallel` value | Effect                                                    |
| -------------------- | --------------------------------------------------------- |
| Not set              | All instances start at the same time                      |
| `1`                  | Instances run one at a time (fully sequential)            |
| `N`                  | At most N instances run simultaneously; others are queued |

{% hint style="info" %}
`max-parallel` applies within a single stage's strategy. It does not affect concurrency between different stages in the pipeline.
{% endhint %}

***

## Next steps

* Go to [Strategy](/harness-platform/3.0/harness-platform-resources/overview/strategy.md) to define matrix, repeat, or parallelism for a stage.
* Go to [Pipelines overview](/harness-platform/3.0/harness-platform-resources/overview.md) to review the pipeline model.