Istio Traffic Metrics
The platform offers a wealth of traffic metrics data, allowing users to analyze service traffic quality from multiple dimensions.
TOC
Prerequisites
The service has been injected with Sidecar, please refer to Adding Services for details.
Quick Start
-
In the left navigation bar, click Monitoring.
-
Select the service you want to view monitoring data for and the time range.
Note: The query time range is limited by the retention period of Prometheus monitoring data. For example: if monitoring data is retained for up to 7 days and the set time range is for the last 30 days, the statistical data will cover only 7 days.
-
Click the respective tabs to view traffic monitoring data and API traffic monitoring data for the service.
Explanation: When the service mesh manages multiple clusters, and there are services with the same namespace and name (non-Dubbo protocol services) in multiple clusters, the traffic monitoring panel displays the aggregated monitoring data of services across multiple clusters.
Use the Cluster Traffic Comparison panel to compare monitoring data of the service in each cluster.
Service Traffic Monitoring
Regular Operations
-
Refresh Data: The monitoring statistics on the current page are automatically refreshed only once when the page is opened. To refresh again, you can use the following two methods:
-
Manual refresh: Click the
in the bottom right corner of the page to manually refresh the data. -
Set auto-refresh (default is off): Click the
to set the interval for auto-refreshing the data.
-
-
View/Set Legend: Click the
in the top right corner of the chart to expand the legend in the monitoring chart. Click on the legend to hide/show the corresponding curve in the chart. -
View Monitoring Chart in Large: Click the
in the top right corner of the chart to view an enlarged, more detailed monitoring chart in the pop-up dialog.
Monitoring Data Explanation
HTTP/HTTP2/gRPC, Dubbo Protocols
| Monitoring Metrics | Description |
|---|---|
| Average Response Time | Average response time within the query time range (total response time/total number of responses). |
| Average Incoming/Outgoing RPS | Average value of incoming/outgoing RPS (Requests Per Second) during the query time range. |
| Response Time | Response time between services or within a service itself, displayed as average, TP 50, TP 95, TP 99. TP (Top Percentile) xx indicates the minimum duration required for xx percent of network requests, commonly used in system performance monitoring scenarios. Hover over the curve to view the response time for a particular period. |
| Incoming/Outgoing RPS | Total incoming/outgoing traffic RPS and error incoming/outgoing traffic RPS within the query time range. RPS = Number of requests during the query time / Query duration (s). |
| Incoming/Outgoing Traffic | Total incoming/outgoing request volume within the query time range, and traffic proportion by HTTP return codes (normal/2xx, 3xx, 4xx, 5xx). Hover over the bar chart to view the traffic proportion for each category. |
| Client Traffic Comparison | Client refers to the client service (downstream service) that makes requests to the current service (upstream service). Client Traffic Comparison will compare and display the Response Time, Incoming RPS, and incoming RPS Error Rate (Error RPS/RPS) of different clients calling the current service. Explanation: - Unknown client refers to the collection of clients (HTTP protocol) within the service mesh that call the current service and whose traffic is not managed by Sidecar. - If the client belongs to the current namespace, clicking the Trace next to the client name will redirect to the trace page. - When there are two versions in a canary release service, aggregate data of both versions is displayed. |
TCP Protocol
| Monitoring Metrics | Description |
|---|---|
| Incoming/Outgoing Traffic | Byte stream size of incoming and outgoing traffic within the query time range. |
| Average Incoming/Outgoing Traffic | Average value of incoming/outgoing traffic (traffic/duration) within the query time range. |
| Incoming/Outgoing | Byte stream transmission rate of incoming and outgoing service network (bytes per second). |
| TCP Connections | Total number of connections. Error Rate = Number of failed connections / Total connections Success Rate = Number of successful connections / Total connections Hover over different colored bar charts to view the number of connections in each category. |
| Client Traffic Comparison | Compares the network incoming byte stream transmission rate between the service and the client services accessing the current service. |
API Traffic Monitoring
To ensure the performance of the monitoring system, you first need to declare service API on the platform. Subsequently, the platform will distinguish the declared API traffic quality in the service's traffic metrics.
After successfully declaring a service API, you can select the API under that service in the API Traffic Monitoring tab.
Monitoring Data Explanation
Tip: During data viewing, you can refer to Regular Operations to refresh monitoring data or hide data corresponding to the legend.
-
Average Response Time: Average response time within the selected time range (total response time/total number of responses).
-
Response Time: Displayed as average, TP 50, TP 95, TP 99. Supports viewing the response time for a single value. Move the mouse over the curve to view the response time for a particular period. TP xx indicates the minimum duration required for xx percent of network requests, commonly used in system performance monitoring scenarios.
-
Average Incoming RPS: Average value of RPS within the selected time range.
-
Incoming RPS: Requests per second, displaying the traffic curve. Supports viewing the total traffic RPS, or click to select the error traffic RPS.
-
Traffic: Displays the total number of visits, average success rate, and average error rate within the selected time range. Colors represent different HTTP status codes: dark green for successful status (2XX), light green for redirection (3XX), orange for client request errors (4XX), and red for server errors (5XX). Average error rate = (4XX+5XX) / total traffic × 100%.