# Monitoring

Reply CMP Monitoring integrates with Grafana to give you near‑real‑time visibility into the health and performance of your cloud resources. Dashboards are grouped by provider and connection so you can drill from a broad overview down to a specific workload.

```{note}
Monitoring shows operational metrics (CPU, memory, latency, requests, etc.). Billing and cost analytics live in {doc}`FinOps <features/finops>`.
```

## Quick start

1) Open Monitoring from the left navigation.  
2) Pick a cloud connection (Azure subscription, AWS account, or GCP project).  
3) Choose a dashboard (for example, Virtual Machines, Databases, Kubernetes).  
4) Use the time range picker and dashboard filters to focus on the resources that matter.

```{tip}
If an empty panel appears, check the time range and that metrics are enabled on the target resource. Some metrics require specific provider features (for example, VM Insights on Azure).
```

## How it works

- Data source: Reply CMP queries provider monitoring APIs (Azure Monitor, AWS CloudWatch, Google Cloud Monitoring) using the service principals tied to your connections.
- Latency: Metrics are near real time; most arrive within a few minutes depending on provider and the metric’s native cadence.
- Security: Access is enforced in Reply CMP and in Grafana. Users only see dashboards and data for their tenant.
- Access options: Dashboards render inside Reply CMP via a secure Grafana embed. You can also open the dashboard directly in Grafana if preferred.

```{note}
Dashboards are customizable on demand. Contact your CMP administrator or your Reply CMP contact point to request new panels, filters, or fully custom dashboards tailored to your needs.
```

## Dashboard selection

From Monitoring:
- Pick a connection (scope).  
- Select a dashboard template for that provider.  
- Adjust filters (resource, namespace, group, region, etc.).  
- Save your favorite time range in the Grafana UI for quick reuse.

---

## Azure dashboards

- Resources Overview: Azure Resource Graph overview across compute, PaaS, networking, monitoring, and security.
- Virtual Machines: Performance and health via Azure Monitor VM Insights.
- App Service Plan: Capacity and runtime metrics for plans and apps.
- Storage Accounts: Performance, capacity, and availability.
- SQL Databases: Health, connections, throughput, and storage.
- Cosmos DB: Throughput, requests, latency, availability, and account management.
- AKS (Kubernetes): Cluster health, nodes, pods, CPU/memory, storage, and networking.

## Amazon Web Services dashboards

- EC2: CPU, network, disk I/O, and instance health.
- RDS: Connections, query performance, storage, and engine‑specific metrics.
- WAFv2: Request patterns, allow/deny rates, and threat signals.
- S3: Bucket counts, storage utilization, request rates, and data transfer.
- Lambda: Invocations, duration, errors, throttles, and concurrency.

## Google Cloud Platform dashboards

- Compute Instances: CPU, memory, disk I/O, network, and instance health.

## Oracle Cloud dashboards

🔜

---

## Data freshness and scope

- Monitoring metrics are operational and typically delayed by only a few minutes.  
- Historical cost corrections (T‑1 backfills) do not affect Monitoring; those belong to {doc}`FinOps <features/finops>` cost pipelines.  
- Some panels aggregate multiple namespaces or services; use the legend or query inspector in Grafana to understand what’s plotted.

## Access and permissions

- You must have access to the connection in Reply CMP.  
- Grafana enforces tenant isolation; you’ll only see your tenant’s data.  
- If you see “unauthorized,” wait a minute after tenant creation or role changes while permissions propagate.

---

## Troubleshooting

### Dashboards don’t load inside the app
Some browsers block third‑party cookies by default, breaking embedded Grafana. Enable third‑party cookies or open the dashboards directly in our Grafana instance: https://grafana.replycmp.com/.

![chrome_enable_3cookies](/_static/images/features/chrome_enable_3cookies.png)

### Unauthorized when opening a dashboard
Right after tenant creation or when your role changes, it can take 1–2 minutes for permissions to propagate. Try again after a short wait.

### Empty or partial graphs
- Expand the time range (for example, Last 6 hours).  
- Verify the connection has permissions to read monitoring APIs.  
- Ensure the resource emits the metric (some require explicit enablement).  
- Use the dashboard’s filters to narrow to a single resource and confirm data flows.

---

## FAQ

**Can I customize dashboards?**  
Yes. Contact your CMP administrator or your Reply CMP contact point to add panels, thresholds, variables, or entirely new dashboards.

**How real‑time are metrics?**  
Typically a few minutes behind live. Exact delay depends on provider and metric granularity.

**How does Monitoring relate to FinOps?**  
Monitoring is for operational telemetry; {doc}`FinOps <features/finops>` is for costs, budgets, and allocations. Jump to FinOps when you need spend analysis.

**Can I open dashboards directly in Grafana?**  
Yes. Use the “open in Grafana” control (where available) or navigate to https://grafana.replycmp.com/ and sign in with your tenant account.