User Guide > Ethernet Port
Ethernet switches connect to each other and to hosts via ethernet ports. This article explains the available settings exposed for Ethernet Ports in the Edit Host and Edit Switch dialog windows. Most attributes are common to both types, however, some attributes are unique to Host Ethernet Ports and Switch Ethernet Ports, respectively.
We begin with discussing some relevant information related to Ethernet layer-2 priority support:
When an Ethernet frame enters an Ethernet Switch via an ingress port, one of the first things that happens is that the ingress interface associates a priority with that particular packet. Depending on the Switch port configuration, the priority is selected from one of the following sources:
VLAN ID: A switch can be configured to give a specific priority to the traffic assigned to a specific VLAN. For example, in an ordinary office network, the network administrator could assign all IP telephony traffic to a dedicated VLAN and give it a higher priority for improved user quality of experience.
Priority based on VLAN ID has precedence over all priority classifications described below.
VLAN tag: For packets carrying a VLAN tag, the packet’s priority can be based on content of the priority bits inside the VLAN tag. The VLAN tag is useful to carry packet priority information on inter-switch links. Use of VLAN tag priority can be configured per port (see below).
IP ToS/DiffServ: For IP packets, the priority can be classified based on the content of the IP ToS bits (IPv4) or the IP TC bits (IPv6). In a traditional office network, classification based on the IP ToS/Diffserv bits could typically be used to provide higher priority to delay sensitive applications, such as IP telephony and remote login, than to bulk data applications such as file transfer. However, a requirement would be that those applications could set the IP ToS/Diffserv bits appropriately. Use of IP ToS/DiffServ priority can be configured per port (see below).
Port Priority: Priority can be classified based on the inbound Switch port. Use of port priority can be configured per port. Furthermore, when priority classification is configured to be based on VLAN tag (or IP ToS/DiffServ), priority will be based on the port priority for untagged (or non-IP respectively) packets.
The priority assigned to a packet will take a value in the range 0-7 and be represented by 3 bits (IEEE 802.1p). If the egress interface of an Ethernet port has, for example, only four traffic class queues, a 802.1p priority (8 values) must be mapped to a traffic class corresponding to one of the four output queues. Different priority to traffic class mappings are used depending on the selected scheduling policy of the port. For the case of four traffic class queues, the default mappings are shown in the tables below. However, the default mappings can be reconfigured, see section Traffic Class Table (TCT) below.
IEEE 802.1P priority | Queue number / Traffic class |
---|---|
0, 1 | 0 (lowest) |
2, 3 | 1 |
4, 5 | 2 |
6, 7 | 3 (highest) |
IEEE 802.1P priority | Queue number / Traffic class |
---|---|
0, 1, 2, 3 | 0 (lowest) |
4 | 2 |
5 | 3 (highest) |
6, 7 | 1 |
IEEE 802.1P priority | Queue number / Traffic class |
---|---|
0, 1, 2, 3, 4 | 0 (lowest) |
5 | 1 |
6 | 2 |
7 | 3 (highest) |
The priority selection of a Switch port is governed by the two fields Port Priority and Port Priority Mode, see below.
After a priority has been associated with a newly arrived Ethernet frame, the frame is then routed to the appropriate output port(s) via an internal switch fabric. In the egress interface of the output port, the priority associated with the Ethernet frame determines into which traffic class queue the Ethernet packet is placed by mapping the Ethernet frame’s priority to a traffic class queue index using the Traffic Class Table (TCT) of the Port.
The available Ethernet port speeds are currently 10 Mbit/s, 100 Mbit/s, 1 Gbit/s and 5 Gbit/s.
Supported scheduling policies are currently Strict Priority Scheduling (SPS), the Credit Based Shaper (CBS) algorithm used by Ethernet AVB switches and Time-Aware Shaper (TAS) for scheduled traffic.
Specifies the (IEEE 802.1p) priority associated with the port. Packets coming in on a specific port will receive this priority unless priority is based on VLAN ID, VLAN tag or IP ToS/DiffServ bits, see Port Priority Mode below.
The port priority mode. Determines if the priority classification for this port is based on content of VLAN tag (IEEE 802.1p priority bits), content of IP ToS/Diffserv bits, or the port priority configured for this port.
VLAN priority settings will have precedence over port priority mode settings.
tag: (Default) The packet’s priority is based on the content of the VLAN tag (802.1p priority bits) of the incoming packet. For packets coming in untagged, the priority is based on the priority associated with the port, see Port Priority above.
ip: The packet’s priority is based on the content of the IP ToS/Diffserv bit of the incoming packet. For non-IP packets coming in on the port (e.g., ARP packets), the priority is based on the priority associated with the port, see Port Priority above.
port: The packet’s priority is based on the priority associated with the port, see Port Priority above.
The number of available traffic class queues in the egress interface of a port.
Specifies the number of stream reservation (SR) classes, which are traffic classes whose bandwidth can be reserved for audio and video (AV) traffic and supports the credit-based shaper algorithm.
A priority value is associated with each SR class. SR classes are denoted by consecutive letters of the alphabet, starting with A and continuing for up to seven classes.
When Scheduling Policy is set to Credit Based Shaper or Time Aware Shaper, SR classes are supported. The maximum number of SR Classes are (Queue Count - 1) and (Queue Count - 2) when Scheduling Policy is set to Credit Based Shaper and Time Aware Shaper, respectively. This field is inactive when the selected Scheduling Policy is Strict Priority Scheduling.
The MAC address assigned to a port.
These are the bandwidth fractions that are reserved for stream reservation (SR) classes.
When the Credit Based Shaper algorithm is the chosen scheduling policy, packets belonging to SR class A and B are currently expected to be assigned priorities 5 and 4, respectively, so you need to specify the reserved bandwidth fraction as a percentage of the total bandwidth for queue number 3 (Fraction A) and queue number 2 (Fraction B) when Queue Count is 4 and queue number 5 (Fraction A) and queue number 4 (Fraction B) when Queue Count is 8.
When the Time-Aware Shaper algorithm is selected as the frame selection transmission algorithm, packets belonging to SR class A and B are currently expected to be assigned priorities 6 and 5, respectively, so you need to specify the reserved bandwidth fraction as a percentage of the total bandwidth for queue number 2 (Fraction A) and queue number 1 (Fraction B) when Queue Count is 4 and queue number 6 (Fraction A) and queue number 5 (Fraction B) when Queue Count is 8.
Selecting this box will activate Time Sensitive Networking (TSN) Frame Preemption on a port.
If frame preemption is active on a Port, a value of frame preemption status is assigned to each value of priority via a frame preemption status table. The possible values of frame preemption status are express or preemptible. When frame preemption is active, you can edit the status table by clicking the edit button. The default value of frame preemption status is express for all priority values.
Each Host and Switch Port provide one or more queues for outbound Ethernet frames, each queue corresponding to a distinct traffic class. Each frame is mapped to a traffic class using the Traffic Class Table for the Port and the frame’s priority. Clicking the edit button allows you to inspect and edit the Traffic class table of a Port. Up to eight traffic classes is supported, allowing separate queues for each priority.
Outbound Ethernet frames mapped to a traffic class greater than or equal to the TCL will be put in the High Traffic Class Frame Buffer Memory of the Port, otherwise they will be put in the Port’s Low Traffic Class Frame Buffer. Default the TCL is set to (QueueCount - 1). See also the discussion in the Switch article.
When an IPDU is to be transmitted encapsulated in a UDP datagram, the UDP datagram is in turn encapsulated in an IPv4 datagram. If the size of the IPv4 datagram exceeds the MTU of the egress Ethernet port, it will be fragmented and sent as a string of two or more Ethernet frames.
Using transmission gates and gate control lists as described in section 8.6.8.4 and onward in [1] and also briefly below, Host and Switch Ports can support enhancements for scheduled traffic, which enables each traffic queue in a Port to schedule transmissions relative to a known timescale.
Especially, when Scheduling Policy is set to Time Aware Shaper (TAS), a transmission gate is associated with each traffic class queue belonging to a Port. The state of the transmission gate for a queue then determines whether or not a frame at the head of the queue is available for transmission:
Furthermore, when TAS is active on a Port, the Port is also associated with one or more gate control lists, each list containing an ordered list of gate operations. Each individual operation updates the state for the transmission gate associated with each of the Port’s traffic class queues allowing it to be set to either Open or Closed. A state is valid for a specified duration until control passes to the next gate operation in the list.
When TAS has been selected as the Scheduling Policy on a port, you can click the Gate Control List (GCL) edit button, which will open the Edit Gate Control Lists dialog. This dialog allows you to define, add and remove gate control lists and specify their individual Base time, the point in time a specific control list is configured and becomes operational. A gate control list that has been configured and is active is called the OperControlList in [1]. The control list that is to be configured next, if any, is referred to as the AdminControlList. Furthermore, within a specific gate control list, the dialog allows you to edit the list of gate operations by adding or removing individual gate states, specify the length or duration of each state as well as the individual state for each transmission gate as Open (o) or Closed (C).
For example, in the figure below, the user has created three gate control lists for a specific port and specified their base times to 760 µs, 2136 µs and 3940 µs, respectively. We see that the last control list consists of four individual gate operations. The sum of their individual durations (or TimeIntervals in [1]) is the CycleTime. In this example it is 495 µs and while a control list is operational, this cycle of gate operations will thus repeat again and again with an interval of CycleTime until it is time to switch in any AdminControlList following the current OperControlList.
Finally, the Time extension parameter allows extending the last gate state in a cycle slightly, if the base time of the next control list to be configured is close in time but not perfectly aligned with the current cycle. This is to avoid restarting the currently operational control list from the beginning and letting it run for just a brief period of time. See section Q.5 on pages 1973-1974 in [1] for a more elaborate motivation and explanation for the use of this parameter.
When opened, this dialog allows you to activate and specify the limits of the Watchdogs that can be associated with a specific Ethernet port.
Especially, the High- and Low Buffer Watchdogs will warn if the corresponding Ethernet-frame memory-buffer of the port ever reaches a fill level exceeding the specified upper limit during a simulation
If activated, the TX watchdog will in a similar manner warn if the utilization of the port transmitter (bandwidth utilization) does not stay below the specified upper bound during a full simulation.
A number of Ethernet packet counters can also be activated to count the occurances of Ethernet packets carrying specific types of payloads, for example, UDP datagrams (more specifically datagram fragments) or different types of PTP or gPTP messages used by the time synchronization protocols. (The difference between PTP and gPTP messages are that the former are encapsulated in UDP datagrams and the latter are carried as the payload of Ethernet frames.)
For each active counter, two watchdogs can also be individually activated. At the end of a simulation, these watchdogs will check that the final counter value is above a lower limit or below an upper limit, respectively.
By selecting the Egress or Ingress tab in an Edit Watchdogs-dialog, counters can be activated individually for the egress and ingress side of an Ethernet port and thus individually inspect the transmitted and received Ethernet packets of that port, respectively.
The set of individual payloads that can be recognized is still limited, but will be expanded in future releases.
For related information, please also see the Simulation results article.
Note! High/Low Buffer are only available on Switch ports.