CloudMonitor displays the status of Server Load Balancer based on seven metrics, including inbound traffic and outbound traffic.  This helps you to monitor the operational status of instances and allows you to configure alarm rules for these metrics. Once you create a Server Load Balancer  instance, CloudMonitor automatically collects data for the following metrics.

Monitoring service

  • Metrics
    • layer-4 metrics
      Metric Definition Dimension Unit Minimum monitoring granularity
      Inbound traffic Traffic consumed by access to the Server Load Balancer from the Internet. Instance  Bps  1 minute
      Outbound traffic Traffic consumed by access to the Internet from the Server Load Balancer. Instance  Bps  1 minute
      Incoming packet count Number of request packets Server Load Balancer receives per second Instance  Count/second 1 minute
      Outgoing packet count Number of request packets Server Load Balancer sends per second Instance  Count/second 1 minute
      New connection count The number of first-time SYN_SENT statuses for TCP three-way handshakes in a statistical period Instance  Count 1 minute
      Active connection count The number of connections in the ESTABLISHED status in the current statistical period Instance  Count 1 minute
      Inactive connection count The number of all TCP connections except connections in the ESTABLISHED status Instance  Count 1 minute
               
               
               1 minute
         The average number of dropped connections per second for a port      
               
               
               
               
               
               
           Instance Count/Second 1 minute
      Inbound instance-discarded packets count Number of inbound packets discarded by the instance per second Instance  Count/Second 1 minute
      Outbound instance-discarded packets count Number of outbound packets discarded by the instance per second Instance  Count/Second 1 minute
      Inbound instance-discarded traffic Amount of inbound traffic discarded by the instance per second Instance bit/s 1 minute
      Outbound instance-discarded traffic Amount of outbound traffic discarded by the instance per second Instance bit/s 1 minute
      Maximum number of concurrent connections for the instance Total number of connections for the instance (sum of the number of active connections and the number of inactive connections) Instance Count/Second 1 minute
      Instance new connections Number of times that the first SYN_SENT state occurs in the case of three TCP handshakes on average per second for the instance in a statistical period Instance Count/Second 1 minute
      Instance inbound packet count Number of request packets received by the instance per second Instance Count/Second 1 minute
      Instance outbound packet count Number of packets sent by the instance per second Instance Count/Second 1 minute
      Instance inbound traffic Traffic consumed by access to the Server Load Balancer instance from the Internet Instance bit/s 1 minute
      Instance outbound traffic Traffic consumed by access to the Internet from the Server Load Balancer instance Instance bit/s 1 minute
    • Layer-7 metrics
      Metric Definition Dimension Unit Minimum monitoring granularity
      Port QPS QPS from listening dimension Port  Count/Second 1 minute
      Port RT Average request delay from port dimension Port  ms 1 minute
      Port status codes 2xx count Number of status codes 2xx the SLB returned to the client from port dimension Port  Count/Second 1 minute
      Port status codes 3xx count Number of status codes 3xx the SLB returned to the client from port dimension Port  Count/Second 1 minute
      Port status codes 4xx count Number of status codes 4xx the SLB returned to the client from port dimension Port  Count/Second 1 minute
      Port status codes 5xx count Number of status codes 5xx the SLB returned to the client from port dimension Port Count/Second 1 minute
      Port other status codes count Number of other status codes the SLB returned to the client from port dimension  Port Count/Second 1 minute
      Port upstream status codes 4xx count Number of status codes 4xx the RS returned to the SLB from port dimension Port  Count/Second 1 minute
      Port upstream status codes 5xx count Number of status codes 5xx the RS returned to the client from port dimension Port  Count/Second 1 minute 
      Port UpstreamRT Average request delay from RS to proxy from port dimension Port  ms 1 minute 
      Instance QPS QPS from instance dimension Instance  Count/Second 1 minute 
      Instance RT Average request delay from instance dimension Instance  Count/Second 1 minute 
      Instance status codes 2xx count Number of status codes 2xx the SLB returned to the client from instance dimension Instance  Count/Second 1 minute 
      Instance status codes 3xx count Number of status codes 3xx the SLB returned to the client from instance dimension Instance  Count/Second 1 minute 
      Instance status codes 4xx count Number of status codes 4xx the SLB returned to the client from instance dimension Instance  Count/Second 1 minute 
      Instance status codes 5xx count Number of status codes 5xx the SLB returned to the client from instance dimension Instance  Count/Second  1 minute 
      Instance other status codes count Number of other status codes the SLB returned to the client from instance dimension Instance  Count/Second 1 minute 
      Instance upstream status codes 4xx count Number of status codes 4xx the RS returned to the SLB from instance dimension Instance  Count/Second 1 minute
      Instance upstream status codes 5xx count Number of status codes 5xx the RS returned to the SLB from instance dimension Instance  Count/Second 1 minute
      Instance upstream RT Average request delay from RS to proxy from instance dimension Instance  ms 1 minute
      Note
      New connection count, active connection count, and inactive connection count all indicate the TCP connection requests from clients to the Server Load Balancer.
  • View metric data
    1. Log on to CloudMonitor console.
    2. Go to the Server Load Balancer instance list under Cloud Service Monitoring.
    3. Click an instance name in the product instance list or click Monitoring Charts from the Actions column to access the instance monitoring details page.
    4. (Optional) Click the Chart Size button to switch to large chart display.

Alarm service

  • Parameter descriptions
    • Metrics: The monitoring indicators provided by Server Load Balancer.

    • Statistical cycle: The alarm system that checks whether your monitoring data has exceeded the alarm threshold value based on the statistical cycle. For example, if the statistical cycle of the alarm rule for memory usage is set to one minute, the system checks whether the memory usage has exceeded the threshold value every other minute.

    • Statistic: This sets the method used to determine if the data exceeds the threshold. You can set Average, Maximum, Minimum, and Sum in Statistic.

      • Average: The average value of metric data within a statistical cycle. The statistical result is the average of all metric data collected within 15 minutes. An average value of over 80% is deemed to exceed the threshold.
      • Maximum: The maximum value of metric data within a statistical cycle. When the maximum value of the metric data collected within the statistical cycle is over 80%, the value exceeds the threshold.
      • Minimum: The minimum value of metric data within a statistical cycle. When the minimum value of the metric data collected within the statistical cycle is larger than 80%, then the value exceeds the threshold.
      • Sum: The sum of metric data within the statistical cycle. When the sum of the metric data collected within the statistical cycle is over 80%, it exceeds the threshold. The preceding statistics are needed for traffic-based indicators.
    • Trigger alarm after the threshold value has exceeded several times: This refers to the alarm that is triggered when the value of the metric continuously exceeds the threshold value in several consecutive statistical cycles.

      For example, you may set the alarm to go off when the CPU usage rate exceeds 80% within a 5-minute statistical cycle after the threshold value is exceeded for three times. If the CPU usage rate is found to exceed 80% for the first time, no warning notification is sent. No alarm is reported if the CPU usage rate exceeds 80% only twice in a row. An alarm is reported only if the CPU usage rate exceeds 80% for the third time. That is, from the first time when the actual data exceeds the threshold to the time when the alarm rule is triggered, the minimum time consumed is the statistical cycle (the quantity of consecutive detection times - 1) = 5 * (3-1) = 10 minutes.

  • Set an alarm rule
    1. Log on to CloudMonitor console.
    2. Go to the Server Load Balancer instance list under Cloud Service Monitoring.
    3. Click Alarm Rules from the Actions column to access the instance’s alarm rules page.
    4. Enter all the relevant information in the required fields, and click Confirm to create a new alarm rule.