Network Interface Stats Telemetry
The network Interface Stats and Global Stats telemetry records are your primary window into the Netify DPI engine's interaction with your network hardware. By providing real-time packet and byte telemetry for every active interface, it allows you to monitor exactly how your traffic is being observed and handled. This visibility ensures that your classification engine is receiving the raw data it needs to maintain high-accuracy visibility.
The interface_stats record provides stats per network interface, while the global_stats record provides the same information summarized across all monitored network interfaces.
Requirements
Interface Stats telemetry was introduced in the Core Plugin in version 1.0.20. To begin streaming the telemetry:
- Install and configure the Netify Agent
- Install and configure a Sink Plugin for receiving the telemetry
- Install and configure the Core Plugin and add stream-stats to types
- Source
- Core Plugin
- Plugin Release
- 1.0.20
Optimizations
The network interface values become significantly more powerful when trended over time to establish a performance baseline. Comparing ingress/egress ratios against dropped and discarded error counters helps teams perform rapid root-cause analysis. Whether you are troubleshooting a misconfigured NIC, a saturated kernel ring buffer, or a processing lag, these statistics provide the empirical evidence needed to resolve bottlenecks before they impact your overall network intelligence and security posture.
- capture_dropped
- A non-zero value indicates that the agent could not service incoming packets quickly enough, resulting in packet loss at the capture layer. If you are using PCAP, this typically means the system requires additional CPU resources. If you are using TPACKET_V3, increase the size of the packet ring buffer to allow the kernel to queue more packets for processing.
- flow_dropped
- A non-zero value indicates that the max_flows limit has been reached. This configuration option controls the maximum number of concurrent flows the Netify Agent tracks, preventing excessive CPU and memory consumption during unusually high traffic volumes or pathological flow patterns. If this counter increases under normal operating conditions, consider raising the max_flows value or scaling the hardware to accommodate the observed traffic load.
Attributes
type
string<interface_name>
objectcapture_dropped
integercapture_filtered
integerdiscarded
integerdiscarded_bytes
integerethernet
integerflow_dropped
integerfragmented
integericmp
integerigmp
integerip
integerip_bytes
integerlargest_bytes
integermpls
integerpppoe
integerqueue_dropped
integerraw
integertcp
integertcp_resets
integertcp_seq_errors
integerudp
integervlan
integerwire_bytes
integerNetwork Interface Stats Attributes - Example
{
"type": "interface_stats",
"eth0": {
"capture_dropped": 0,
"capture_filtered": 0,
"discarded": 12,
"discarded_bytes": 1012,
"ethernet": 772,
"flow_dropped": 0,
"fragmented": 0,
"icmp": 3,
"igmp": 2,
"ip": 760,
"ip_bytes": 261408,
"largest_bytes": 1514,
"mpls": 0,
"pppoe": 0,
"queue_dropped": 0,
"raw": 772,
"tcp": 648,
"tcp_resets": 0,
"tcp_seq_errors": 0,
"udp": 107,
"vlan": 0,
"wire_bytes": 279648
}
}
Global Stats Attributes - Example
{
"capture_dropped": 0,
"capture_filtered": 0,
"discarded": 13,
"discarded_bytes": 1070,
"ethernet": 2049,
"flow_dropped": 0,
"fragmented": 0,
"icmp": 4,
"igmp": 13,
"ip": 2036,
"ip_bytes": 1218047,
"largest_bytes": 4369,
"mpls": 0,
"pppoe": 0,
"queue_dropped": 0,
"raw": 2049,
"tcp": 1844,
"tcp_resets": 4,
"tcp_seq_errors": 0,
"type": "global_stats",
"udp": 175,
"vlan": 0,
"wire_bytes": 1266911
}