Alibaba Cloud DNS:TCP health checks

Overview

TCP health check is a network protocol used by GTM to perform health checks on target addresses. It primarily monitors metrics such as network reachability, port availability, and latency of IP addresses. When an IP address or port becomes abnormal, the system automatically blocks the abnormal IP address. When the IP address returns to normal, the system automatically removes the block.

Configuration method

1. Log on to Alibaba Cloud DNS - Global Traffic Manager.

2. On the instance list page of Global Traffic Manager, click the Actions column of the target instance, and then click the Configuration button.

3. On the Basic Configuration page, click the Address Pool tab, then click the "+" icon in front of the address pool, and then click Add next to Health Check.

4. In the Modify Health Check dialog box, complete the configuration of Health Check Protocol, Health Check Interval, Health Check Port, Timeout Period, Consecutive Failures, Failure Ratio, and Monitoring Nodes, and then click OK.

   

Parameters

1. Health Check Protocol

   Select TCP to monitor the target IP address using the TCP protocol. This primarily monitors metrics such as network reachability, port availability, and latency of the IP address.

2. Health Check Interval

   Select the time interval for each TCP monitoring. By default, a check is performed every 1 minute. The minimum health check interval supported is 15 seconds (available for Ultimate Edition users).

3. Health Check Port

   Checks whether Telnet can be performed on this port of the target IP address. If Telnet is successful on the port, the service is considered normal. If Telnet fails on the port, the service is considered abnormal.

4. Timeout Period

   For each TCP health check, the system calculates the return time of the sent TCP packet. If the packet does not return within the timeout period, the health check is considered timed out. The available timeout values are: 2 seconds, 3 seconds, 5 seconds, and 10 seconds.

5. Consecutive Failures

   When performing TCP health checks, the system determines that the application service is abnormal only after multiple consecutive abnormal detections. This prevents monitoring accuracy from being affected by factors such as momentary network jitter. The available values for consecutive failures are: 1 time, 2 times, and 3 times.

• 1 time: The application service is determined to be abnormal when the health check detects one alarm.

• 2 times: The application service is determined to be abnormal when the health check detects two consecutive alarms.

• 3 times: The application service is determined to be abnormal when the health check detects three consecutive alarms.

6. Failure Ratio

   When performing TCP health checks, this is the ratio of abnormal monitoring points to the total number of monitoring points. When the failure ratio exceeds the set threshold, the application service is determined to be abnormal. The available failure ratio thresholds are: 20%, 50%, 80%, and 100%.

7. Monitoring Nodes

   The geographic location of the node that performs TCP health checks. The default monitoring nodes provided by the system are as follows: