Page 151 / 998 Scroll up to view Page 146 - 150
Layer 2 Switching in the Network
Overview
2-1
iMG/RG Software Reference Manual (Switching)
2. Switching
2.1
Overview
2.1.1
Layer 2 Switching in the Network
The System consists of a Layer 2 switch coupled to a Network Processor. The aggregate is viewable as a single
Layer 2 switch, but this functionality is spread across the two devices - switch and the bridge - with interconnec-
tivity being provided by the CPU port.
Rate Limiting, QOS - and VLAN Tag management is provided at the edge of the system
- via
port configura-
tion.
By default - all traffic flows in one single VLAN - however an extension to this model is to use VLANs to segre-
gate traffic flows to certain ports.
2.1.2
Documentation Structure
The Preface listed all of the iMG/RG/iBG devices and to which product category they belong. Keeping this in
mind, the user can better use the remainder of this section, which is organized as follows:
An overview of an area and its main attributes.
The functions within an area. These are explained in some detail, usually with accompanying figures.
A table that lists these functions and to which product category they apply. Notes help the user understand
why a function may or may not be relevant.
A table that lists the commands and to which product category they apply.
A command reference for each command and its parameters.
Note:
The command reference subsection is generic for all product categories. The user should refer
to the the function and command tables to see how a command or parameter applies to a
specific product.
2.2
Switching
2.2.1
Overview
The iMG/RG/iBG product includes an integrated layer 2 managed switch providing Fast Ethernet transceivers
supporting 10Base-T, 100Base-TX and 1000Base-TX modes, high performance memory bandwidth (wire speed)
and an extensive feature set including Rate Limiting, QoS priority, VLAN tagging and MIB counters.
Page 152 / 998
Switching
Layer 2 switch functional description
iMG/RG Software Reference Manual (Switching)
2-2
The layer 2 switch uses one additional l 00Mbps or 1000Mbps port as an internal port to communicate to the
central processor in order to access layer 3 services such as routing, VoIP protocols, firewall and NAT security
modules.
The following is the complete set of features available in the switch module:
IEEE 802.1q tag based VLAN (up to 16 VLANs)
VLAN ID tag/untag options, per port basis
Programmable rate limiting, ingress port, egress port, per port basis.
IGMP v1/v2 snooping for multicast packet filtering
QoS packet prioritization support: per port, IEEE 802.1p and DiffServ based
Integrated look-up engine with dedicated 1 K unicast MAC addresses
Automatic address learning, address aging and address migration
Full duplex IEEE 802. flow control
Automatic MDI/MDI-X crossover for plug-and-play on all the ports
2.2.2
Layer 2 switch functional description
A summary of the general switch functions is included below.
2.2.2.1 Port Management
All ports on the switch are numbered sequentially from “lan1” up to the max number of Lan based 10/100
Ethernet ports. For the available number, please see the summary table in the preface. There can be special
function LAN interfaces - such as HPNA - that are addressed where that function is discussed. The admin sta-
tus of the port can be set - as well as the
Port Status and Counter value being displayed.
The port speed can also be set - as one of the following options: 100MFull, 100MHalf, 10MFull, 10MHalf, Auto,
Coax. The Coax mode is used when connecting an Ethernet to Coax Balun to the device.
2.2.2.2 Ingress Filtering
The infiltering parameter enables or disables Ingress Filtering of frames admitted on the ports.
If a port has only TAGGED VLANs associated with it - then when InFiltering is set to:
ON - Only TAGGED packets with a VLAN ID matching VLANs associated with the port are admitted.
UNTAGGED Packets are not admitted.
OFF - Both TAGGED packets with a VLAN ID Matching VLANS associated with the port are admitted - as
well as UNTAGGED packets. UNTAGGED Packets are tagged with the Default VLAN ID.
Page 153 / 998
Layer 2 switch functional description
Switching
2-3
iMG/RG Software Reference Manual (Switching)
2.2.2.3 Address management
The primary function of the layer 2 switch is to receive good packets from the ports, process them and forward
them to the appropriate ports for transmission. This frame processing involves the Ingress
Policy, Queue Con-
troller, Output Queues and Egress Policy.
The normal packet flow involves learning how to switch packets only to the correct ports. The switch learns
which port and end station is connected to by remembering each packet's Source Address along with the port
number on which the packet arrived - and the vlan that it is on.
When a packet is directed to a new, unlearned MAC address, the packet is flooded out of all the ports (as long
as they belong on the same VLAN) except for the one on which it arrived. Once a MAC address/port number is
learned, all future packets directed to that end station's MAC addresses are directed to the learned port num-
ber only. This ensures that the packet is sent to the correct end station. This table can be displayed via the CLI
The address database is stored in the embedded switch memory and has a default aging time of about 300 sec-
onds (5 minutes). If no packets are received from that MAC Address during that aging interval, then the address
is purged from the database. If a MAC Address is received from a different port during this time, then the MAC
address is learned on that new port and all traffic is then routed to that new port.
The number of MAC addresses that can be learned differs between devices. (Kendin, BCM, Marvell, Marvell Gig)
2.2.2.4 Rate limiting support
The integrated layer 2 switch supports hardware rate limiting on receive and transmit independently on a per
port basis. The rate limiting applies to all the frame types: unicast, broadcast
and multicast.
Some devices do provide the ability to rate limit the Multicast and Broadcast traffic. (BCM and Gig Marvell)
If the number of bytes exceeds the programmed limit, the switch will stop receiving or transmitting packets on
the port. In the transmit direction, extra packets are placed in one or more FIFO queues and sent as soon as
possible given the configured limit. Note that when multiple queues are configured, the highest priority queue is
emptied first.
In the receive direction, on some devices, there is an option provided for flow control to prevent packet loss. In
this case, if the configured limit is reached, and Flow Control is enabled, then a PAUSE frame will be sent to the
peer device. This will stop transmission of packets until the Gateway is ready to receive packets again.
2.2.2.5 Loop Detection
Loop detection is a feature available at layer 2 used to disable automatically one or more switch ports when a
loop is verified on one or more of these ports.
Ethernet loops are likely to happen when a Ethernet-to-Coax balun is used in installations where there are
appliances connected to coax cable that need to the6 ethernet ports. In this case, if the coax cable is not prop-
erly terminated, a signal reflection is generated on the coax cable segment and then reported to the ethernet
segment too causing high network degradation.
Page 154 / 998
Switching
Layer 2 switch functional description
iMG/RG Software Reference Manual (Switching)
2-4
To detect a loop on ethernet ports, the Gateway periodically sends a “special” ping message. If the gateway
receives the same ping message back, it means that a loop is present. In this case the Gateway disables all the
traffic to/from the port (except the “special” ping) until the loop has been removed.
2.2.2.6 Layer 3 Routing Rate Limiting
The integrated layer 2 switch can limit traffic that goes to the Gateway network processor where routing tasks
need to be performed.
Limitation on the maximum routing rate is necessary to preserve system resources for high priority tasks like
VoIP and IGMP.
If the number of frames per seconds that need to be routed to the network processor are higher than the
selected maximum rate, the layer 2 switch discards packets addressed to the network processor in order to
force the average traffic rate to be below the target rate.
2.2.2.7 Quality of Service Classification
QoS switching policy is performed by the Queue Controller. The priority of a frame is determined in priority
order by:
The IEEE 802.3ac Tag containing IEEE 802.1p priority information: this IEEE 802.1p priority information is
used in determining frame priority when IEEE 802.3ac tagging is enabled on the port.
The IPv4 Type of Service (TOS)/DiffServ field when enabled on the port. IPv4 priority classification can be
configured on a port basis to have a higher priority then IEEE Tag.
The user can enable these classification individually or in combination.
All untagged frames entering a port have their priority set to the port's default priority. This priority is then
used to manage the traffic from that port.
There are two different models in place:
1.
A two Queue scheme- where by the user specifies which Priority settings go into the high priority queue and
which go into the low queue.
2.
A four Queue scheme where the user actually maps the different priority values to one of the four queues.
Highest priority queues are emptied first before the lower priority queues…and as such, it is possible for the
low priority traffic to get starved out.
The integrated layer 2 switch supports two
Class of Service
(CoS) mechanisms:
IEEE 802.1p
tagging (Layer 2)
and
Differentiated Services
(DS) as an advanced architecture of ToS (Layer 3).
2.2.2.7.1 802.1p traffic priority
The IEEE 802.1p signalling technique is an IEEE endorsed specification for prioritizing network traffic at the
data-link/MAC sub-layer (OSI Reference Model Layer 2).
Page 155 / 998
Layer 2 switch functional description
Switching
2-5
iMG/RG Software Reference Manual (Switching)
IEEE 802.1p
is a spin-off of the
IEEE 802.1q (VLAN tagging) standard and they work in tandem (see Figure 1).
The 802.1q standard specifies a VLAN tag that appends to a MAC frame. The VLAN tag carries VLAN informa-
tion. The VLAN tag has two parts: The VLAN ID (12-bit) and User Priority (3-bit). The User Priority field was
never defined in the VLAN standard. The 802.1q implementation defines this prioritizing field.
Switches, routers, servers, even desktop systems, can set these priority bits in the three-bit user priority field,
which allows packets to be grouped into various traffic classes. If a packet is received that does not have this tag
added, then the switch adds it to the packet and uses the default priority associated with the port.
In the two queue systems, the user priority field in the TAG header is compared with an internal value inthe
switch called the base priority - and all values equal or greater to this base priority are put into the high priority
egress queue - while all others are put into the low priority queue.
In the four queue systems, the value in the user priority is used to determine which queue to place the packet
into directly. This mapping is configurable.
2.2.2.7.2 Differentiated services code point (DSCP)
The IEEE 802.1p signalling technique is an IEEE endorsed specification for prioritizing network traffic.
The DSCP octet in the IP header classifies the packet service level. The DSCP replaces the ToS Octet in the
IPv4 header (see
Figure 2-1
).
Currently, only the first six bits are used. Two bits of the DSCP are reserved for future definitions. This allows
up to 64 different classifications for service levels.
In the two queue systems, the DSCP field is compared with an internal value in the switch called the base prior-
ity - and all values equal or greater to this base priority are put into the high priority egress queue - while all
others are put into the low priority queue.
In the four queue systems, the value in the user priority used to determine which queue to place the packet into
directly. This mapping is configurable.

Rate

4 / 5 based on 3 votes.

Popular Allied-Telesis Models

Bookmark Our Site

Press Ctrl + D to add this site to your favorites!

Share
Top