Monitoring Metrics

Overview

This section describes the namespace, the metrics that can be monitored by Cloud Eye, and dimensions of these metrics. You can view the metrics reported by ELB and the generated alarms on the Cloud Eye console. For details, see Viewing Metrics.

Namespace

SYS.ELB

Metrics

Table 1 Metrics supported by ELB

Metric ID

Name

Description

Value

Monitored Object

Monitoring Period (Raw Data)

m1_cps

Concurrent Connections

Load balancing at Layer 4: total number of TCP and UDP connections from the monitored object to backend servers

Load balancing at Layer 7: total number of TCP connections from the clients to the monitored object

Unit: N/A

>= 0

  • Load balancer

  • Listener

1 minute

m2_act_conn

Active Connections

Number of TCP and UDP connections in the ESTABLISHED state between the monitored object and backend servers

You can run the following command to view the connections (both Windows and Linux servers):

netstat -an

Unit: N/A

>= 0

m3_inact_conn

Inactive Connections

Number of TCP connections between the monitored object and backend servers except those in the ESTABLISHED state

You can run the following command to view the connections (both Windows and Linux servers):

netstat -an

Unit: N/A

>= 0

m4_ncps

New Connections

Number of TCP and UDP connections established between clients and the monitored object per second

Unit: Count/s

>= 0/second

m5_in_pps

Incoming Packets

Number of packets received by the monitored object per second

Unit: Packet/s

>= 0/second

m6_out_pps

Outgoing Packets

Number of packets sent from the monitored object per second

Unit: Packet/s

>= 0/second

m7_in_Bps

Inbound Rate

Traffic used for accessing the monitored object from the Internet per second

Unit: byte/s

>= 0 bytes/s

m8_out_Bps

Outbound Rate

Traffic used by the monitored object to access the Internet per second

Unit: byte/s

>= 0 bytes/s

m9_abnormal_servers

Unhealthy Servers

Number of unhealthy backend servers associated with the monitored object

Unit: N/A

>= 0

  • Load balancer

1 minute

ma_normal_servers

Healthy Servers

Number of healthy backend servers associated with the monitored object

Unit: N/A

>= 0

mb_l7_qps

Layer-7 Query Rate

Number of requests the monitored object receives per second

Unit: Query/s

>= 0 query/s

  • Load balancer

  • Listener

1 minute

md_l7_http_3xx

3xx Status Codes

Number of 3xx status codes returned by the monitored object

Unit: Count/s

>= 0/second

  • Load balancer

  • Listener

1 minute

mc_l7_http_2xx

2xx Status Codes

Number of 2xx status codes returned by the monitored object

Unit: Count/s

>= 0/second

  • Load balancer

  • Listener

1 minute

me_l7_http_4xx

4xx Status Codes

Number of 4xx status codes returned by the monitored object

Unit: Count/s

>= 0/second

mf_l7_http_5xx

5xx Status Codes

Number of 5xx status codes returned by the monitored object

Unit: Count/s

>= 0/second

m10_l7_http_other_status

Other Status Codes

Number of status codes returned by the monitored object except 2xx, 3xx, 4xx, and 5xx status codes

Unit: Count/s

>= 0/second

m11_l7_http_404

404 Not Found

Number of 404 Not Found status codes returned by the monitored object

Unit: Count/s

>= 0/second

m12_l7_http_499

499 Client Closed Request

Number of 499 Client Closed Request status codes returned by the monitored object

Unit: Count/s

>= 0/second

m13_l7_http_502

502 Bad Gateway

Number of 502 Bad Gateway status codes returned by the monitored object

Unit: Count/s

>= 0/second

m14_l7_rt

Average Layer-7 Response Time

Average response time of the monitored object

The response time starts when the monitored object receives requests from the clients and ends when it returns all responses to the clients.

Unit: ms

>= 0 ms

m1a_l7_upstream_rt_max

Maximum Server Response Time

Maximum response time of backend servers (This metric is available only when the frontend protocol is HTTP or HTTPS.)

The response time starts when the monitored object routes the requests to the backend server and ends when the monitored object receives a response from the backend server.

Unit: ms

>= 0 ms

  • Load balancer

  • Listener

1 minute

m1b_l7_upstream_rt_min

Minimum Server Response Time

Minimum response time of backend servers (This metric is available only when the frontend protocol is HTTP or HTTPS.)

The response time starts when the monitored object routes the requests to the backend server and ends when the monitored object receives a response from the backend server.

Unit: ms

>= 0 ms

m1c_l7_rt_max

Maximum Layer-7 Response Time

Maximum response time of the monitored object (This metric is available only when the frontend protocol is HTTP or HTTPS.)

The response time starts when the monitored object receives requests from the clients and ends when it returns all responses to the clients.

Unit: ms

>= 0 ms

  • Load balancer

  • Listener

1 minute

m1d_l7_rt_min

Minimum Layer-7 Response Time

Minimum response time of the monitored object (This metric is available only when the frontend protocol is HTTP or HTTPS.)

The response time starts when the monitored object receives requests from the clients and ends when it returns all responses to the clients.

Unit: ms

>= 0 ms

l7_con_usage

Layer-7 Concurrent Connection Usage

Ratio of HTTP and HTTPS connections established between the monitored object and backend servers per second, to the maximum number of concurrent connections allowed per second

Unit: percent (%)

>= 0%

Load balancer

1 minute

l7_in_bps_usage

Layer-7 Inbound Bandwidth Usage

Ratio of the bandwidth that the monitored object uses to return response to clients over HTTP and HTTPS, to the maximum outbound bandwidth allowed

Unit: percent (%)

Inbound bandwidth usage <100 %

Caution

CAUTION: If the inbound bandwidth usage reaches 100%, the load balancer performance has reached the upper limit. If the inbound bandwidth keeps higher than the bandwidth that the load balancer can provide, the service availability cannot be guaranteed.

>= 0%

l7_out_bps_usage

Layer-7 Outbound Bandwidth Usage

Ratio of the bandwidth that the monitored object uses to return response to clients over HTTP and HTTPS, to the maximum outbound bandwidth allowed

Unit: percent (%)

Outbound bandwidth usage <100 %

Caution

CAUTION: If the outbound bandwidth usage reaches 100%, the load balancer performance has reached the upper limit. If the outbound bandwidth keeps higher than the bandwidth that the load balancer can provide, the service availability cannot be guaranteed.

>= 0%

l7_ncps_usage

Layer-7 New Connection Usage

Ratio of HTTP and HTTPS connections established between clients and the monitored object per second, to the maximum number of new connections allowed per second

Unit: percent (%)

>= 0%

l7_qps_usage

Layer 7 QPS Usage

Ratio of HTTP and HTTPS queries per second on the monitored object, to the maximum number of queries allowed per second

Unit: percent (%)

>= 0%

l4_con_usage

Layer-4 Concurrent Connection Usage

Ratio of TCP and UDP connections established between the monitored object and backend servers per second, to the maximum number of concurrent connections allowed per second

Unit: percent (%)

>= 0%

Load balancer

1 minute

l4_in_bps_usage

Layer-4 Inbound Bandwidth Usage

Ratio of the bandwidth that the monitored object uses to receive requests from clients over TCP and UDP, to the maximum inbound bandwidth allowed

Unit: percent (%)

Inbound bandwidth usage <100 %

Caution

CAUTION: If the inbound bandwidth usage reaches 100%, the load balancer performance has reached the upper limit. If the inbound bandwidth keeps higher than the bandwidth that the load balancer can provide, the service availability cannot be guaranteed.

>= 0%

l4_out_bps_usage

Layer-4 Outbound Bandwidth Usage

Ratio of the bandwidth that the monitored object uses to return response to clients over TCP and UDP, to the maximum outbound bandwidth allowed

Unit: percent (%)

Outbound bandwidth usage <100 %

Caution

CAUTION: If the outbound bandwidth usage reaches 100%, the load balancer performance has reached the upper limit. If the outbound bandwidth keeps higher than the bandwidth that the load balancer can provide, the service availability cannot be guaranteed.

>= 0%

l4_ncps_usage

Layer-4 New Connection Usage

Ratio of TCP and UDP connections established between clients and the monitored object per second, to the maximum number of new connections allowed per second

Unit: percent (%)

>= 0%