---
title: "Services"
description: "The Services page in Curade Mini grades each service on its response time and failure rate, and shows its throughput, its hosts and its problems."
url: https://curade.ai/en/curade-mini/docs/features/services/
product: "Curade Mini"
version: "1.0.0"
breadcrumbs:
  - "Curade Mini docs"
  - "Features"
  - "Services"
---

# Services

The Services page in Curade Mini grades each service on its response time and failure rate, and shows its throughput, its hosts and its problems.

**Services** covers the services behind the applications: how fast they respond, how often they fail and how much traffic they carry. Engineers open it; Admins do not.

![Service detail with response time, throughput and failure rate over a Cluemetry card naming the root cause and a deployment](https://curade.ai/_astro/05-service-cluemetry.CrmxNdfr_1kqLXS.webp "The why behind every number.")

## The list

Two tabs split the list, **Services** and **Database**, since the two are read differently. Each card carries the service's name and type, how many hosts it runs on, a grade worked out from its response time and failure rate over your window, and a badge saying whether it was active, inactive or reported no data.

- Inactive means the service handled no requests in the window although a host it runs on was up. No data means it reported nothing while its hosts were down or not reporting.
- A service that reported nothing has nothing to grade, so it is badged **Not graded** rather than healthy.
- A service that needs attention carries one plain line saying why, such as too many requests failing or no requests in the window.
- Search, **Add Filter** and **Sort By** work on both tabs, and the sort covers the three metrics and the number of hosts.

## The service's page

1. The header: name, grade, type, technology, management zones and tags, with **Copy details**.
2. The headline numbers over your window, with a line naming what they cover.
3. **What's behind these numbers** ([Cluemetry](https://curade.ai/en/curade-mini/docs/features/cluemetry/)).
4. **Runs on host**, with **View host**, and **Change host** when the service runs on several hosts.
5. The relationship map: what calls the service, every host it runs on and everything it calls ([Relationship map](https://curade.ai/en/curade-mini/docs/features/relationship-map/)).
6. The problems raised against the service inside the window.
7. Charts of its request numbers, and of the CPU, memory and disk of the host in view.

## Reading the numbers

| Number | What it means |
|---|---|
| Response time | The average time a request took over the window. A long window smooths a spike out of sight, so narrow it to see whether something is recent. |
| Failure rate | The share of requests that failed. |
| Throughput | Requests per minute. |

All three are counted per request, so a host that served ten thousand requests weighs far more than one that served a dozen. That is how a service can read calm across its hosts while one of them struggles: on a service with several hosts, pick the struggling host with **Change host** to read it alone, where the environment records request numbers per host.

The list grades a service across all its hosts, while the page grades the host in view, so a service healthy in the list can open as critical on one host. The page says so when it happens.