This article discusses non-flow metrics used in Kentik queries.
Non-flow metrics allow you to monitor health indicators, such as CPU, Memory, and Interface utilization, that traditional flow records do not capture. These metrics are primarily derived from SNMP polling and Streaming Telemetry (ST).
Notes:
For general information about how Kentik handles data, see About Metrics.
For metrics specific to individual hardware vendors, see Device-Specific Metrics.
SNMP Metrics
When SNMP OID Polling is enabled, Kentik enriches the Kentik Data Engine (KDE) with health and performance data. These metrics use Universal Data Records (UDR), providing a flexible schema that adapts to different device types.
Query Restrictions
To ensure data accuracy, SNMP metrics are subject to the following logic in the Dimension Selector:
No Mixing: SNMP metrics cannot be mixed with flow-based metrics (e.g., Bits/s) in the same query.
Dimension Matching: Results only return if the query’s dimensions (Group By/Filter) are also from the SNMP category.
SNMP Interface Metrics
These metrics are collected from network interfaces using standard Kentik-Polled SNMP OIDs. They are typically used to monitor port saturation and physical link health.
Metric in Portal | Variations | OID | Description |
|---|---|---|---|
Bit Rate | Input, Output |
| Rate of bits received/transmitted, including framing. |
Packets | Input, Output |
| Rate of unicast packets (excludes discards). |
Errors | Input, Output |
| Inbound/outbound packets containing delivery-preventing errors. |
Discards | Input, Output |
| Packets dropped by the device (e.g., due to buffer exhaustion). |
Multicast | Input, Output |
| Rate of packets addressed to a multicast group. |
Broadcast | Input, Output |
| Rate of packets addressed to a broadcast address. |
TIP: All SNMP metrics above can be aggregated in the Portal as Average, 95th Percentile, Max, or Total.
SNMP Device Metrics
Device metrics track the internal health of a chassis or its individual components (e.g., line cards). These are identified in the portal using the Component dimension.
Metric in Portal | Description |
|---|---|
CPU | The percentage of CPU utilization for the component. |
Memory Total | The total physical memory (in bytes) on the component. |
Memory Used | The amount of memory (in bytes) currently in use. |
Memory Free | The amount of memory (in bytes) currently available. |
Memory Utilization | The percentage of total memory used by the component. |
For a list of supported OIDs by manufacturer, see Vendor-Specific SNMP OIDs.
Streaming Telemetry Metrics
If Streaming Telemetry is enabled, you can query interface metrics that are pushed by the device in real-time rather than polled via SNMP.
IMPORTANT: Which ST are you using?
NMS Implementation: If you are using Kentik NMS, refer to NMS via Streaming Telemetry.
Legacy Implementation: This section describes the non-NMS Streaming Telemetry workflow used for classic KDE queries.
Usage in Portal
Streaming Telemetry metrics share the same names and calculation methods as the SNMP Interface Metrics listed above. However, when building a query (e.g., in Data Explorer), you must select them from the Streaming Telemetry category in the Metrics dialog.
Note: Just like SNMP, ST metrics cannot be mixed with metrics from other categories in a single query.