---
title: "Monitor a Load Balancer"
description: "Watch the traffic a TIR Load Balancer serves, from requests and connections to latency, errors, and healthy backends, over the last 5 minutes to 12 hours."
content_type: "how-to"
product: "tir-load-balancer"
platform: "TIR"
url: "/tir/tir-load-balancer/how-to/monitor-a-load-balancer/"
topic: ["monitor"]
---

# Monitor a Load Balancer

Open the **Monitoring** tab of a Load Balancer to see the traffic it is serving as charts: how many
requests arrive, how fast they are answered, how many fail, and how many backends are healthy.

## Before you begin

- A Load Balancer that has finished creating. See [Create a Load
  Balancer](/tir/tir-load-balancer/how-to/create-a-load-balancer/).
- The read permission on the Load Balancer service. See [TIR Access Control](/tir/tir-iam/).

## Open the charts

1. In the side menu, under **Compute**, select **Load Balancer**, then select the Load Balancer's name.
2. Select the **Monitoring** tab.
3. Choose a time range: **5m**, **15m**, **30m**, **1h**, **3h**, **6h**, or **12h**. The range starts at
   **1h**.

Each chart has a point every 15 seconds, and its times are in Indian Standard Time. Select the refresh
button to reload every chart.

## Read the charts

Each metric has its own chart with a name and a description. **Data In** and **Data Out** share one
**Traffic** chart.

| Chart                  | What it shows                                                                                                                                   |
| ---------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- |
| **Requests / sec**     | Requests received per second.                                                                                                                   |
| **Rate Limited / sec** | Requests per second refused for going over the limit of the plan.                                                                               |
| **Throttle Rate**      | The percentage of requests refused for going over the limit of the plan.                                                                        |
| **Active Connections** | Client connections open now.                                                                                                                    |
| **Data In**            | Bytes per second received from clients.                                                                                                         |
| **Data Out**           | Bytes per second sent to clients.                                                                                                               |
| **Latency P90**        | The time, in milliseconds, within which 90% of requests are served.                                                                             |
| **Avg Response Time**  | The average time to serve a request, in milliseconds.                                                                                           |
| **Error Rate**         | The percentage of requests that reached a backend and got a 5xx answer.                                                                         |
| **Success Rate**       | The percentage of requests that got a 2xx answer. Requests refused for going over the limit are left out of both this chart and **Error Rate**. |
| **CPU Utilization**    | The Load Balancer's CPU use as a percentage of one vCPU.                                                                                        |
| **Healthy Backends**   | The number of backends that pass their health check, added up over all backend groups.                                                          |

Byte values read as B, KB, MB, GB, or TB per second, and percentage charts run from 0 to 100.

A chart with an empty window says "No data in this window". Charts that depend on backends, such as
**Healthy Backends**, get data once the Load Balancer has a backend; until then they stay empty
rather than showing zero. If the tab says "Failed to fetch metrics", select the refresh button.

To be warned when a figure goes wrong, [set up alerts](/tir/tir-load-balancer/how-to/set-up-alerts/).
To read what the Load Balancer itself reports, see [View logs](/tir/tir-load-balancer/how-to/view-logs/).
For help with empty charts, see [Troubleshooting TIR Load Balancer](/tir/tir-load-balancer/troubleshooting/tir-load-balancer/).