Network Topology metrics reference

This document describes Cloud Monitoring metrics and metadata, which you can view in a Network Topology visualization for each entity and connection.

Metadata describes data such as name, description, and location.
Metrics describe data such as traffic, latency, and packet loss, which are the supported metrics for connections.

For each Network Topology hierarchy, the Google Cloud Console displays a single metric for Compute Engine virtual machine (VM) instance entities and region entities, as well as for connections. You can change this default metric.

For more information, see the Network Topology overview and the Graph interface overview.

For steps describing how to work with metrics and metadata, see Using Network Topology.

Metrics for VM or regional entities

This section lists metrics for VMs or regional entities.

Entity	Metrics
VM	CPU utilization Memory usage percentage Disk usage percentage
Region	Traffic

Metrics for connections between entities

This section lists metrics for connections between entities.

Some connections support packet loss or latency metrics. Packet loss is based on probers that estimate the packet loss that you might experience, while latency and other metrics are based on your own traffic.

If there are not enough probes or traffic to calculate a high confidence value for packet loss or latency, Network Topology displays an asterisk (*) next to the value on the graph. For more information about how these metrics are calculated, see the Metrics section of the Performance Dashboard overview.

Connections in the VM hierarchy

This section lists metrics for connections in the aggregation hierarchy where VM is the base entity. For more information about entities and their aggregation hierarchies, see Resources and traffic.

Connection	Metrics
VM to VM¹	Traffic Latency: based on `round-trip time (RTT)` Packet loss²

¹ VM-to-VM flows are not available at all aggregation levels. For example, the VM-to-VM flow is not available, but the VM-to-region flow is available.

² Packet loss is available only for regions and zones.

Client to load balancing

This section lists metrics and metadata for traffic from external or internal clients to different types of Cloud Load Balancing.

Entity (node or edge)	Metrics
External client³ to HTTP(S) Load Balancing	Traffic: based on `request bytes count` and `response bytes count` Latency: based on `Frontend RTT` Request count HTTP error response count: based on `responses by response code class` HTTP error response percent: based on `response fraction by code class` Client-side error rate (total): based on CSM metrics
External client to SSL Proxy Load Balancing or TCP Proxy Load Balancing	Traffic: based on `inbound traffic` and `outbound traffic` Latency: based on `Frontend RTT` Current open connections: based on `open connections` Connections open rate: based on `new connections per second` Connections close rate: based on `closed Connections per second`
External client to Network Load Balancing	Traffic: based on `external bytes count` for egress and ingress

³For external clients, Network Topology supports only aggregation levels from Country and higher.

Load balancing to backend instance groups

This section lists metrics and metadata from different types of Cloud Load Balancing to instance group backends.

Entity or connection	Metrics
HTTP(S) Load Balancing to backend (instance group)	Traffic: based on `backend request bytes count` and `backend response bytes count` Latency: based on `backend latencies` Request Count: based on `backend request count`
SSL Proxy Load Balancing or TCP Proxy Load Balancing to backend (instance group)	Traffic: based on `inbound traffic` and `outbound traffic`
Internal TCP/UDP Load Balancing to backend (instance group)	Traffic: based on `inbound traffic` and `outbound traffic`
Network Load Balancing to backend (instance group)	Traffic: based on `external bytes count` for egress and ingress

VLAN attachments

This section lists metrics and metadata for the data transferred between your VPC network and the on-premises network, as counted by the VLAN attachment. An on-premises network can refer to any remote network that is outside the Google Cloud domain. The connection between Google Cloud and on-premises can traverse either dedicated or partner Interconnect devices.

Entity or connection	Metrics
VPC network to the on-premises network	Traffic: based on `egress bytes count` and `ingress bytes count`

VPN tunnels

This section lists metrics and metadata for the data transferred between your VPC network and the on-premises network, as counted by the VPN tunnel.

Entity or connection	Metrics
VPC network to the on-premises network	Traffic: based on `egress bytes count` and `ingress bytes count`

VM to VLAN attachments

This section lists metrics and metadata for the data transferred between your VM and VLAN attachment. The connection between Google Cloud and on-premises can traverse either dedicated or partner Interconnect devices.

Entity or connection	Metrics
VM to VLAN attachments	Traffic: based on `egress bytes count` and `ingress bytes count`

VM to VPN tunnels

This section lists metrics and metadata for the data transferred between your VM and VPN tunnels.

Entity or connection	Metrics
VM to VPN tunnels	Traffic: based on `egress bytes count` and `ingress bytes count`

Client to Google services

This section lists metrics and metadata for traffic from internal clients to different types of Google-managed services.

Entity or connection	Metrics
VPC network to Google-managed services	Traffic: based on `egress bytes count` and `ingress bytes count`