# TiDB Cloud Cluster Monitoring Metrics

__*TiDB Cloud Cluster Monitoring Metrics*__

| Metrics | Description |
| :-- | :-- |
| Total QPS | Count the number of SQL statements executed per second on all TiDB instances by SELECT, INSERT, UPDATE, etc. Numerical results are related to application load |
| Latency | The time when the client network request is sent to TiDB and return to the client after TiDB execution is finished. In general, client request are sent in the form of SQL statements, but they can also include the execution time of commands such as COM*PING, COM*SLEEP, COM*STMT*FETCH, COM*SEND*LONG_DATA, etc. |
| Connections | Number of connections per TiDB |
| TiFlash (Beta) Request Duration | The total time for all TiFlash instances to process a coprocessor request. The total time is from when the coprocessor request is received until the request is answered |
| TiFlash (Beta) Storage Size | The amount of storage space used by each TiFlash instance |
| TiKV Storage Size | The amount of storage used by each TiKV instance |
| TiDB CPU | CPU usage per TiDB instance |
| TiKV CPU | CPU usage per TiKV instance |
| TiKV IO Read (bytes/s) | Disk read throughput per TiKV instance |
| TiKV IO Write (bytes/s) | Disk write throughput per TiKV instance |

-------------------------------------------------------------
On the cluster overview page, you can view commonly used metrics for the cluster.

    Navigate to the Active Clusters page.

    Click the cluster name to go to the cluster overview page.

    Currently, these metrics include Total QPS, Latency, Connections, TiFlash Request QPS, TiFlash Request Duration, TiFlash Storage Size, TiKV Storage Size, TiDB CPU, TiKV CPU, TiKV IO Reads, and TiKV IO Writes.


Total QPS
Count the number of SQL statements executed per second on all TiDB instances by type SELECT, INSERT, UPDATE, etc. The numerical result is related to the application load.
- Observe whether QPS has stabilized in the last cycle, and whether there is a significant increase or decrease compared to the previous cycle.
- Observe whether the QPS of each TiDB instance is balanced.

Latency
The time when the client network request is sent to TiDB and return to the client after TiDB execution is finished. In general, client request are sent as SQL statements, but they can also include the execution time of commands such as COM*PING, COM*SLEEP, COM*STMT*FETCH, COM*SEND*LONG_DATA, and so on.
Since TiDB supports Multi-Query, the client can send multiple SQL statements at once, such as select 1; select 1; select 1;. The execution time at this point is the total time since all SQL statements have been executed.
- Observe whether the trend is stable within the same period and whether the values are balanced between different instances.

Connections
The number of connections per TiDB.
- Observe whether the number of connections has stabilized in the last cycle, and whether there is a significant increase or decrease compared to the previous cycle.
- Observe whether the number of connections to each TiDB instance is balanced.

TiFlash (Beta) Request Duration
The total time for all TiFlash instances to process a coprocessor request. The total time is from the time the coprocessor request is received to the time the request is answered.

TiFlash (Beta) Storage Size
The amount of storage space used by each TiFlash instance.
- Observe the amount of used storage for each instance, and consider scaling up if it accounts for 80% of the available capacity.

TiKV Storage Size
The amount of storage used by each TiKV instance.
- Observe the amount of used storage for each instance, and consider scaling up if it accounts for 80% of the available capacity.

TiDB CPU
The CPU usage of each TiDB instance.
- Observe the CPU usage of each instance. If it occupies 80% of the available capacity for a long time, you need to consider scale out.
- Observe whether the trend is stable within the same period and whether the values are balanced between different instances.

TiKV CPU
The CPU usage of each TiKV instance.
- Observe the CPU usage of each instance. If it occupies 80% of the available capacity for a long time, you need to consider scale out.
- Observe whether the trend is stable within the same period and whether the values are balanced between different instances.

TiKV IO Read (bytes/s)
Disk read throughput per TiKV instance
- Observe whether the trend is stable within the same period and whether the values are balanced between different instances.

TiKV IO Write (bytes/s)
Disk write throughput per TiKV instance
- Observe whether the trend is stable within the same period and whether the values are balanced between different instances.