Advanced Topics

The SGSW-2403 supports Layer 2 switching and other advanced features, which are described in this chapter.

When a frame enters a port, its destination MAC address is checked in the address database to see which port leads to this destination. If the destination address belongs to the incoming port, the frame is dropped or "filtered" because it addressed to the local segment. If the destination address if found on another port, the frame is forwarded to that port and queued for output. But, if the destination address is not found in the address database, the frame is sent to one or more output ports based on the rules for handling tagged or untagged VLAN frames.

If the source MAC address of the frame was not found in the address database, it is recorded along with the incoming port number where it entered the switch. This information is then used to make later decisions for frame forwarding.

The Spanning Tree Algorithm (that is, the STA configuration algorithm as outlined in IEEE 802.1D) can be used to detect and disable network loops, and to provide link backup. This allows the switch to interact with other bridging devices(including STA compliant switches, bridges or routers) in your network to ensure that only one route exists between any two stations on the network. If redundant paths or loops are detected, one or more ports are put into a blocking state (stopped from forwarding packets) to eliminate the extra paths. Moreover, if one or more of the paths in a stable spanning tree topology fail, this algorithm will automatically change ports from blocking state to forwarding state to reestablish contact with all network stations.

The STA uses a distributed algorithm to select a bridging device (STA compliant switch, bridge or router) that serves as the root of the spanning tree network. It selects a root port on each bridging device (except for the root device) which incurs the lowest path cost when forwarding a packet from that device to the root device. Then it selects a designated bridging device from each LAN which incurs the lowest path cost when forwarding a packet from that LAN to the root device. All ports connected to designated bridging devices are assigned as designated ports. After determining the lowest cost spanning tree, it enables all root ports and designated ports, and disables all other ports. Network packets are therefore only forwarded between root ports and designated ports, eliminating any possible network loops.

Once a stable network topology has been established, all bridges listen for Hello BPDUs (Bridge Protocol Data Units) transmitted from the Root Bridge. If a bridge does not get a Hello BPDU after a predefined interval (Maximum Age), the bridge assumes that the link to the Root Bridge is down. This bridge will then initiate negotiations with other bridges to reconfigure the network to reestablish a valid network topology.

The following figure gives an illustration of how the Spanning Tree Algorithm assigns bridging device ports.

Switches do not inherently support broadcast domains, which can lead to broadcast storms in large networks that handle a lot of IPX or NetBeui traffic. In conventional networks with routers, broadcast traffic is split up into separate domains to confine broadcast traffic to the originating group and provide a much cleaner network environment. Instead of using physically separate subnets which are linked by traditionally slow routers, this switch creates segregated broadcast domains based on easily configurable VLANs, and then links these VLANs as required with wire-speed routing.

An IEEE 802.1Q VLAN is a group of ports that can be located anywhere in the network, but communicate as though they belong to the same physical segment. VLANs help to simplify network management by allowing you to move devices to a new VLAN without having to change any physical connections. VLANs can be easily organized to reflect departmental groups (such as Marketing or R&D), usage groups (such as e-mail), or multicast groups (used for multimedia applications such as video conferencing).

VLANs provide greater network efficiency by reducing broadcast traffic, but also allow you to make network changes without having to update IP addresses or IP subnets. VLANs inherently provide a high level of network security, since traffic must pass through a Layer 3 switch or a router to reach a different VLAN.

Before enabling VLANs for the switch, you must first assign each port to the VLAN group(s) it will participate in. (By default all ports are assigned to VLAN 1 as untagged ports.) Add a port as a tagged port (that is, a port attached to a VLAN-aware device) if you want it to carry traffic for one or more VLANs and the device at the other end of the link also supports VLANs. Then assign the port at the other end of the link to the same VLAN(s). However, if you want a port on this switch to participate in one or more VLANs, but the device at the other end of the link does not support VLANs, then you must add this port as an untagged port (that is, a port attached to a VLAN-unaware device).

Port-based VLANs are tied to specific ports. The switch's forwarding decision is based on the destination MAC address and its associated port. Therefore, to make valid forwarding and flooding decisions, the switch learns the relationship of the MAC address to its related port--and thus to the VLAN--at run-time. When the switch receives a frame, it assigns the frame to the port's default VLAN if the frame is untagged (determined by the PVID of the receiving port), or maps it for output to the broadcast domain associated with the frame's VLAN tag.

Port overlapping can be used to allow access to commonly shared network resources among different VLAN groups, such as file servers or printers. Note that if you implement VLANs which do not overlap, but still need to communicate, you can connect them using the Layer-3 router/switch.

GVRP defines a system whereby the switch can automatically learn the VLANs each endstation should be assigned to. If an endstation (or its network adapter) supports the IEEE 802.1Q VLAN protocol, it can be configured to broadcast a message to your network indicating the VLAN groups it wants to join. When this switch receives these messages, it will automatically place the receiving port in the specified VLANs, and then forward the message to all other ports. When the message arrives at another switch that supports GVRP, it will also place the receiving port in the specified VLANs, and pass the message on to all other ports.
VLAN requirements are propagated in this way throughout the network. This allows

GVRP-compliant devices to be automatically configured for VLAN groups based solely on endstation requests.

This switch only supports 255 VLANs, but up to 4094 VLANs are supported by the IEEE 802.1Q VLAN protocol. Therefore, if this switch is attached to endstations that issue VLAN registration requests, it will have to forward unknown VLAN tags. This traffic can only be propagated to the rest of the network if automatic VLAN registration is enabled on your switch.

Ports can be assigned to multiple tagged or untagged VLANs. Each port on the switch is therefore capable of passing tagged or untagged frames. To forward a frame from a VLAN-aware device to a VLAN-unaware device, the switch first decides where to forward the frame, and then strips off the VLAN tag. However, to forward a frame from a VLAN-unaware device to a VLAN-aware device, the switch first decides where to forward the frame, and then inserts a VLAN tag reflecting this port´s default VID. The default PVID is VLAN 1 for all ports, but this can be changed (see page 2-41 or page 3-28).

The switch supports intra-VLAN communication using wire-speed switching. However, if you have devices in separate VLANs that must communicate, and it is not practical to include these devices in a common VLAN, then the VLANs can be connected via the Layer-3 switch (such as the ES3616) or router.

Multicasting sends data to a group of nodes instead of a single destination. The simplest way to implement multicasting is to broadcast data to all nodes on the network. However, such an approach wastes a lot of bandwidth if the target group is small compared to overall the broadcast domain.

Since applications such as video conferencing and data sharing are more widely used today, efficient multicasting has become vital. A common approach is to use a group registration protocol that lets nodes join or leave multicast groups. A switch or router can then easily determine which ports contain group members and send data out to those ports only. This procedure is called multicast filtering.

The purpose of multicast filtering is to optimize a switched network's performance, so multicast packets will only be forwarded to those ports containing multicast group hosts or multicast routers/switches instead of flooding to all ports in the subnet (VLAN). The Switch support multicast filtering by passively monitoring IGMP Query and Report messages.

A layer 2 switch can passively snoop on IGMP Query and Report packets transferred between IP Multicast Routers/Switches and IP Multicast host groups to learn the IP Multicast group members. It simply monitors the IGMP packets passing through it, picks out the group registration information, and configures multicast filters accordingly. IGMP Snooping generates no additional network traffic, allowing you to significantly reduce the multicast traffic passing through your switch.

The Internet Group Management Protocol (IGMP) runs between hosts and their immediately neighboring multicast router/switch. IGMP is a multicast host registration protocol that allows any host to inform its local router that it wants to receive transmissions addressed to a specific multicast group.

A router, or multicast-enabled switch, can periodically ask their hosts is they want to receive multicast traffic. If there is more than one router/switch on the LAN performing IP multicasting, one of these devices is elected "querier" and assumes the responsibility of querying the LAN for group members.

Based on the group membership information learned from IGMP, a router/switch can determine which (if any) multicast traffic needs to be forwarded to each of its ports. At Layer 3, multicast routers use this information, along with a multicast routing protocol such as DVMRP, to support IP multicasting across the Internet.

Note that IGMP neither alters nor routes any IP multicast packets. A multicast router/ switch must be used to deliver IP multicast packets across different subnetworks.

The SGSW-2403 provides two transmit queues on each port, with a Weighted Fair Queuing scheme. This function can be used to provide independent priorities for various types of data such as real-time video or voice, and best-effort data.

Priority assignment to a packet in the SGSW-2403 can be accomplished in any of the following ways:

Ports can be combined into an aggregate link to increase the bandwidth of a network connection or ensure fault recovery. You can group ports into trunks that consist of two, three or four ports, creating an aggregate bandwidth of up to 8 Gbps when grouping multiple Gigabit ports. Besides balancing the load across each port in the trunk, the additional ports provide redundancy by taking over the load if another port in the trunk should fail.

SNMP (Simple Network Management Protocol) is a communication protocol designed specifically for managing devices or other elements on a network. Network equipment commonly managed with SNMP includes hubs, switches, bridges, routers and host computers. SNMP is typically used to configure these devices for proper operation in a network environment, as well as monitor them to evaluate performance and detect potential problems.

Remote Monitoring (RMON) provides a cost-effective way to monitor large networks by placing embedded or external probes on distributed network equipment (hubs, switches or routers). SNMP network management software like castle Rock's SNMPc can access the probes embedded in recent PLANET network products to perform traffic analysis, troubleshoot network problems, evaluate historical trends, or implement proactive management policies. RMON has already become a valuable tool for network managers faced with a quickly changing network landscape that contains dozens or hundreds of separate segments. RMON is the only way to retain control of the network and analyze applications running at multi-megabit speeds. It provides the tools you need to implement either reactive or proactive policies that can keep your network running based on real-time access to key statistical information.

This switch provides support for mini-RMON which contains the four key groups required for basic remote monitoring. These groups include:

Statistics: Includes all the tools needed to monitor your network for common errors and overall traffic rates. Information is provided on bandwidth utilization, peak utilization, packet types, errors and collisions, as well as the distribution of packet sizes.

History: Can be used to create a record of network utilization, packet types, errors and collisions. You need a historical record of activity to be able to track down intermittent problems. Historical data can also be used to establish normal baseline activity, which may reveal problems associated with high traffic levels, broadcast storms, or other unusual events. Historical information can also be used to predict network growth and plan for expansion before your network becomes too overloaded.

Alarms: Can be set to test data over any specified time interval, and can monitor absolute or changing values (such as a statistical counter reaching a specific value, or a statistic changing by a certain amount over the set interval). Alarms can be set to respond to either rising or falling thresholds.

Events: Defines the action to take when an alarm is triggered. The response to an alarm can include recording the alarm in the Log Table or sending a message to a trap manager. Note that the Alarm and Event Groups are used together to record important events or immediately respond to critical network problems.

[Top of this Chapter]