Overview… statistics, packet counts and rates
All traffic leaving and entering each test port is counted, and there are separate counts for a variety of different traffic classes: per port, per stream, per test payload, per filter, etc.
The basic statistics for each traffic class are:
- Percentage, of the physical bandwidth
- Mbits/sec, mega-bits in the last second
- Packets/sec, packets in the last second
- Bytes, accumulated number of bytes
- Packets, accumulated number of packets
The numbers can be quite long, and a “,” is used to separate the digits into groups of three. A “.” is the decimal separator.
Whereas percentages are measured at layer-1, Mbits/sec and Bytes values are at layer-2. So they include only the useful portions of an Ethernet packet, from and including the destination MAC address up to and including the frame check sum, and exclude the preamble and any additional inter frame gap.
Observed Mbits/sec rates will therefore always be below the physical rate of the port which is specified at layer-1 and hence includes everything. For instance, with a packet length of 80 bytes and an inter frame gap of 20 bytes, 100% load on a 10 Gbps link will result in 8000 Mbps of traffic at layer-2.
In the XenaManager, the statistics are grouped into transmit statistics and receive statistics. Each group has some buttons:
- “Mark”, draws each current value in a faint color, making it easy to see when the value has changed.
- “Clear”, clears all counts to zero. This requires the port to be reserved.
- “Save”, saves the current values to a file, for instance for logging or post-processing purposes.
On the transmit side, there is a set of statistics for each stream generated by the port:
On the receive side, the central statistics concern the various test payloads that have been observed in the incoming packets. They are counted separately for each TID value:
The “Clean” button erases the list of observed TIDs. It is rebuilt as packets arrive with test payloads containing new TID values.
In addition to the traffic counters, there is a second section showing the error checks, and a third section showing latency measurements for each test payload stream. Some of these require the source port for the stream to be identified.
The receive side also has a section showing the traffic counts for each filter:
