Token metrics
The Tokens per second (TPS) metric reports the tokens generated by the device per second. The number of tokens generated may seem backwards at first glance because, a trunk with many devices, passes fewer tokens per second. The fewer the devices on the trunk, the more tokens generated.
There is no rule-of-thumb for tokens per second, no good or bad. Tokens per second varies wildly based on device count, vendor implementation, baud rate, trunk utilization, and more. This limits the use of the metric to viewing changes over time. You need to establish a baseline for each particular trunk, and then investigate any deviations from its unique norm.
The following is an example of a couple of inflection points on a graph.
These are the types of problems that this graph can help you investigate.
- Notice that from 7 am until 2:45 pm the tokens per second indicates a healthy MS/TP trunk, passing tokens at a constant rate. It is impossible to see the effect of applications messages at this level, as repetitive applications messages cause a repetitive impact on TPS.
- Each spike or deviation was the result of adding or removing devices to or from the trunk. When the TPS initially dropped to zero (0) shortly after 3 pm, the controller completely stopped passing tokens. In this case, the MS/TP connector was unplugged from the controller.
- The steady increase to ~13 TPS occurred when two other devices were removed from the trunk.
- Shortly thereafter, the two devices were restored to the network and the TPS returned to its previous level (8).
- The TPS again dropped to zero (0) when the controller was again unplugged.
- Once the controller was plugged back in again, things returned to the normal 8 TPS. (The angle of the line is a COV (Coefficient of Variation) history artifact, the value was zero (0) until it was five (5).
This sequence of events represents only the tip of the iceberg of possibilities. When Parse All Properties is set to true, the Tokens component exposes all of the current properties and any that may be added in the future.