Monitoring Network Load Balancing

This document provides you with the information you need to understand Cloud Monitoring for Network Load Balancing.

Network Load Balancing exports key metrics to Monitoring.

Monitoring metrics can be used for the following purposes:

  • Evaluate a load balancer's configuration, usage, and performance
  • Troubleshoot problems
  • Improve resource utilization and user experience

You can create custom dashboards, set up alerts, and query metrics through Monitoring.

Configuring monitoring and alerting using Cloud Console

Refer to Metrics and resource types for a list of collected metrics. Refer to Filters for lists of attributes you can use to filter your results.

Defining Monitoring custom dashboards

You can create custom Monitoring dashboards over Network Load Balancing metrics.

  1. Go to Monitoring in the Google Cloud Console.
    Go to Monitoring
  2. Select Dashboards > Create Dashboard.
  3. Click Add chart.
  4. Give the chart a title.
  5. Select metrics and filters.
    You can find Network Load Balancing metrics by searching for the loadbalancing.googleapis.com/l3/external prefix.
  6. Click Save.

Defining Monitoring alerts

You can define Monitoring alerts over various Network Load Balancing metrics:

  1. Go to Monitoring in the Google Cloud Console.
    Go to Monitoring
  2. Select Alerting > Create a Policy.
  3. Select resource types and metrics.
    You can find Network Load Balancing resource types and metrics by searching for loadbalancing.googleapis.com/l3/external in the Find resource type and metric field.
  4. Specify a Configuration to set when an alert should be triggered.
  5. Add Filters if desired.
  6. Click Save.

Metric reporting frequency and retention

Metrics for the network load balancers are exported to Monitoring in 1-minute granularity batches. Monitoring data is retained for six weeks. The dashboard provides data analysis in default intervals of 1h (one hour), 6h (six hours), 1d (one day), 1w (one week), and 6w (six weeks). You can manually request analysis in any interval from six weeks (6w) to sixty seconds (60s).

Metrics and resource types

Metrics

The following metrics for network load balancers are reported into Monitoring. You can use these metric names when making API requests.

metric_name Type Description
loadbalancing.googleapis.com/l3/external/ingress_bytes_count counter The number of bytes sent from a client to a network load balancer backend. For TCP flows, only bytes in the application stream are counted. Sampled every 60 seconds. After sampling, data is not visible for up to 150 seconds.
loadbalancing.googleapis.com/l3/external/ingress_packets_count counter The number of packets sent from a client to a network load balancer backend. Sampled every 60 seconds. After sampling, data is not visible for up to 150 seconds.
loadbalancing.googleapis.com/l3/external/egress_bytes_count counter The number of bytes sent from a network load balancer backend to a client. For TCP flows, only bytes in on the application stream are counted. Sampled every 60 seconds. After sampling, data is not visible for up to 150 seconds.
loadbalancing.googleapis.com/l3/external/egress_packets_count counter The number of packets sent from a network load balancer backend to a client. Sampled every 60 seconds. After sampling, data is not visible for up to 150 seconds.
loadbalancing.googleapis.com/l3/external/rtt_latencies

Available only for TCP traffic.		 distribution A distribution of RTT measured over TCP connections for network load balancer flows. Sampled every 60 seconds. After sampling, data is not visible for up to 150 seconds.

Resource types

resource.type = tcp_lb_rule or udp_lb_rule

A network load balancer can support either TCP or UDP traffic. You can specify the resource type to limit data to either TCP or UDP load balancers.

Filters

Resource labels

You can restrict your query by resource. You can also group results by these values.

resource.label.<var>label_key</var>:
label_key Type Description
project string The identifier of the Google Cloud project associated with this resource, such as "my-project".
load_balancer_name string The name of the load balancer.
region string The region where the load balancer's backend instance is located: us-central1, europe-west1, asia-east1
network_name string The VPC network in which the load balancer backend resides.
backend_target_type string The type of the backend target that handled the connection. The only value is "target_pool."
backend_target_name string The name of the load balancer's target pool.
forwarding_rule_name string The name of the forwarding rule used by the client to connect to the network load balancer.
forwarding_rule_network_tier string The network tier of the forwarding rule.
backend_name string The name of the backend that handled the connection.
backend_type string The type of the backend that handled the connection.
backend_scope string The scope (zone or region) of the backend group that handled the connection. This field is unsupported for target pools. For a network load balancer, the field would always say UNSUPPORTED_FOR_TARGET_POOL.
backend_scope_type string The type of the scope of the backend group that handled the connection. Valid values are "zone" and "region." This field is unsupported for target pools. For a network load balancer, the field would always say UNSUPPORTED_FOR_TARGET_POOL.
backend_failover_configuration string The configured state of the target pool that handled the connection.
backend_subnetwork_name string The name of the subnetwork of the instance that received the connection.
endpoint_zone string Zone of the instance that handled the connection.

Metric labels

You can restrict your query by metric label. You can also group results by these values.

metric.label.<var>label_key</var>:
label_key Type Description
client_country string Country of the client which initiated the connection to the network load balancer.
client_continent string Continent of the client which initiated the connection to the network load balancer.

Monitoring API requests

You can construct arbitrary queries over the Network Load Balancing metrics using the Monitoring API`s v3 projects.timeSeries.list request. The exact structure of the various definitions follow the generic monitoring filters semantics.

Example API requests

  1. Get all bytes sent from all network load balancers in the project, during 12:00-12:02 PM UTC on 2019-07-01, with 1m aggregates:

timeSeries.list parameters:

  • name: projects/PROJECT_ID
  • Filter: resource.type = "tcp_lb_rule" AND metric.name="loadbalancing.googleapis.com/l3/external/egress_bytes_count"
  • interval.start_time: 2019-07-01T12:00:00Z
  • interval.end_time: 2019-07-01T12:02:00Z
  • aggregation.alignmentPeriod: 60s
  • aggregation.crossSeriesReducer: REDUCE_SUM
  • aggregation.perSeriesAligner: ALIGN_SUM
  1. Get median RTT measurement over all network load balancers in the project, during 12:19-12:20 PM UTC on 2019-07-01, broken down by client country:

timeSeries.list parameters:

  • name: projects/PROJECT_ID
  • Filter: resource.type = "tcp_lb_rule" AND metric.name="loadbalancing.googleapis.com/l3/external/rtt_latencies"
  • interval.start_time: 2019-07-01T12:19:00Z
  • interval.end_time: 2019-07-01T12:20:00Z
  • aggregation.alignmentPeriod: 60s
  • aggregation.crossSeriesReducer: REDUCE_SUM
  • aggregation.groupByFields: metric.label.client_country
  • aggregation.perSeriesAligner: ALIGN_PERCENTILE_50
  1. Get total bytes from a specific network load balancer during 12:19-12:20 PM UTC on 2019-07-01, broken down by endpoint_zone and client continent:

timeSeries.list parameters:

  • name: projects/PROJECT_ID
  • Filter: resource.type = "tcp_lb_rule" AND resource.label.load_balancer_name = "netlb-bs-1" AND metric.name="loadbalancing.googleapis.com/l3/external/ingress_bytes_count"
  • interval.start_time: 2019-07-01T12:19:00Z
  • interval.end_time: 2017-07-01T12:20:00Z
  • aggregation.alignmentPeriod: 60s
  • aggregation.crossSeriesReducer: REDUCE_SUM
  • aggregation.groupByFields: metric.label.client_continent
  • aggregation.groupByFields: resource.label.endpoint_zone
  • aggregation.perSeriesAligner: ALIGN_SUM

What's next