---
title: Metrics
description: Chart a service's CPU, memory, network, disk, request rate and response time over the last hour to the last seven days.
section: Observability
order: 2
---

The project's **Metrics** tab charts a service's CPU and memory, and its network, disk, requests and response time when it has them. Pick a service and a **Time window** of 1 hour, 6 hours, 24 hours or 7 days. Measurements start a minute or two after a service is running.

## Which series are there?

| Series | Chart | Unit |
| --- | --- | --- |
| `cpu_percent` | CPU | Percent |
| `memory_mib` | Memory, against the service's limit | MiB |
| `network_in_bytes` | Network in | Bytes per step |
| `network_out_bytes` | Network out | Bytes per step |
| `disk_mib` | Disk | MiB |
| `requests_per_minute` | Requests | Requests per minute |
| `p95_latency_ms` | Response time, 95th percentile | Milliseconds |

## How far apart are the points?

| Window | Step |
| --- | --- |
| `1h` | 60 seconds |
| `6h` | 5 minutes |
| `24h` | 15 minutes |
| `7d` | 1 hour |

## Read metrics from the API

```bash tab="cURL"
curl "https://api.dekka.cloud/v1/resources/$SERVICE_ID/metrics?window=24h" \
  -H "Authorization: Bearer $DEKKA_API_KEY"
```

The response holds `step_seconds` and a list of `series`, each with a `name`, a `unit` and its `points`. While the dashboard shows the **Mock engine** label, `simulated` is `true` and the numbers are made up.