Lindorm Insight allows you to view the performance monitoring data of clusters. This provides an easy method for you to view the resource usage of clusters. The Cluster Monitor page displays the monitoring data of cluster running from multiple dimensions, such as IDC and group.

Go to the Cluster Monitor page

Log on to the Lindorm Insight system. In the left-side navigation pane, choose Monitor > Cluster Monitor. Then, the Cluster Monitor page in Lindorm Insight appears. You can filter and view cluster monitoring data by IDC, group, and time range.

Cluster monitoring
The Cluster Monitor page consists of two parts: Cluster overview and Overview of QP layer. These two parts are divided into multiple sections. You can click the following links for the sections to view the corresponding data:

Server Load

The Server Load section displays the average load of the servers in your cluster.Server Load
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.Load
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

CPU Usage

The CPU Usage section displays the average CPU usage of the servers in your cluster.CPU Usage
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.cpuUser
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

bytesIn

The bytesIn section displays the inbound traffic of your cluster.bytesIn
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.bytesIn(MB/s)
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

bytesOut

The bytesOut section displays the outbound traffic of your cluster.bytesOut(MB/s)
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.bytesOut(MB/s)
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Overall read throughput of the cluster

The Overall read throughput of the cluster section displays the total number of rows scanned by the bottom layer when your cluster runs queries, such as select, wget, and wscan. This section displays the overall read load metric that collects the data of various query methods. This metric corresponds to the read request metric for Lindorm servers in the original monitoring system.Overall read throughput of the cluster
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.readRequest
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Response time of cluster overall query

The Response time of cluster overall query section displays the average time that your cluster consumes to read each row. You can use the following formula to calculate the response time: Response time = The total time of cluster queries per unit time/The number of rows scanned by the bottom layer. This metric is an overall metric that describes the average time of reading each row. If you need to query the metric data of specific query types, check the WScan, WGet, and other request details in the following sections.Response time of cluster overall query
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.readRT
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Overall write throughput of the cluster

The Overall write throughput of the cluster section displays the total number of rows scanned by the bottom layer when your cluster performs operations to modify data. The examples of the operations include upsert, wput, and wdelete. This section displays the overall write load metric that collects the data of various query methods.Overall write throughput of the cluster
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.writeRequest
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Response time of the cluster overall write

The Response time of cluster overall write section displays the average time that your cluster consumes to write each row. You can use the following formula to calculate the response time: Response time = The total time of cluster write operations per unit time/The number of rows scanned by the bottom layer. This metric is an overall metric that describes the average time of writing each row. If you need to query the metric data of specific query types, check the WPut, WDelete, and other request details in the following sections.Response time of the cluster overall write
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.writeRT
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Cluster WGet details

The Cluster WGet details section displays the performance metric data of the get commands that your cluster runs. The following list describes the metric data:
  • ops: the number of requests per second
  • AVG: the average response time of requests
  • MAX: the maximum response time of requests
  • rowsPerOp: the average number of rows affected by each operation
  • 50%RT: the median response time of requests
  • 99.9%RT: the 99.9% response time of requests
Cluster WGet details
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.ops
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Cluster WPut details

The Cluster WPut details section displays the performance metric data of the put commands that your cluster runs.Cluster WPut details
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.AVG
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Cluster WScan details

The Cluster WScan details section displays the performance metric data of the Scan commands that your cluster runs.Cluster WScan details
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.ops
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Cluster WDelete details

The Cluster WDelete details section displays the performance metric data of the Delete commands that your cluster runs.Cluster WDelete details
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.ops
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

Cluster Select details

The Cluster Select details section displays the performance metrics data of the Select commands that your cluster runs.Cluster Select details
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.AVG
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.

QP upsert details

The QP upsert details section displays the performance metric data of the Upsert commands that your cluster runs.QP upsert details
To view the metric data of a server, double-click the metric curve. Then, the metric data that corresponds to the list of servers in your cluster appears.AVG
Note By default, the top 10 servers for the metric are displayed. If the number of servers is excessively large, you can click the server name to show or hide the corresponding metric curve.