Smart Link & Monitor Link Technology White Paper

Keywords: Smart Link, Monitor Link, control VLAN, flush message, rapid switchover, dual uplink, link redundancy

Abstract: Smart Link is a feature developed to provide effective, reliable link redundancy, load sharing, and fast convergence for dual-uplink networks. Monitor Link is a feature developed to complement the link backup mechanism of Smart Link. By monitoring the uplink, and synchronizing the downlink with the uplink, Monitor Link triggers the switchover between the primary and backup links in a smart link group. This document mainly describes the basic concepts, mechanisms, and typical application scenarios of Smart Link and Monitor Link.

Acronyms:

Acronym

Full spelling

RRPP

Rapid Ring Protection Protocol

STP

Spanning Tree Protocol

 

 



Smart Link Overview

1.1  Background

Dual uplink networks are widely used nowadays. As shown in Figure 1 , Switch A is dually uplinked to Switch D through Switch B and Switch C.

Figure 1  Diagram of a dual uplink network

Although dual uplinks can provide link redundancy in the network, the possible loop (Switch A — Switch B — Switch D — Switch C — Switch A) can also cause broadcast storms. Hence, you need to take measures to prevent loops in the network. In most cases, STP is used to eliminate network loops. The problem with STP, however, is that STP convergence time is long, and a large amount of traffic is lost during the convergence time. Therefore, STP is not suitable for convergence delay-sensitive application environments. RRPP is another effective solution used to remove loops. It has shorter convergence time, but it is mostly used in complex ring-shaped networks. Besides, RRPP is hard to configure. To address this problem, H3C developed the Smart Link solution.

1.2  Benefits

Smart Link is dedicated to dual-uplink networks. It delivers the following benefits:

l              Keeping one uplink connected and the other blocked when both uplinks in a dual uplink network are healthy, thus preventing broadcast storms caused by network loops.

l              Switching the traffic to the backup link within a few subseconds when the primary link fails, thus ensuring the normal forwarding of traffic in the network.

l              Easy to configure.

Smart Link Implementation

2.1  Basic Concepts in Smart Link

2.1.1  Smart Link Group

A smart link group consists of two member ports: the master port and the slave port. A port can belong to different smart link groups at the same time. Normally, at a time, only one port (master or slave) is active for forwarding, while the other port is blocked, that is, in the standby state. When link failure occurs on the active port due to port shutdown or Ethernet OAM link errors for example, the standby port becomes active to take over while the former active port transits to the blocked state.

Figure 2  Smart Link application scenario

As shown in Figure 2 , GigabitEthernet 1/0/1 and GigabitEthernet 1/0/2 of Switch D form a smart link group (marked in blue). GigabitEthernet 1/0/1 is in the forwarding state (marked by a continuous line), and GigabitEthernet 1/0/2 is in the blocked state (marked by a broken line). GigabitEthernet 1/0/1 and GigabitEthernet 1/0/2 of Switch E form another smart link group (marked in red). GigabitEthernet 1/0/1 is in the blocked state (marked by a broken line), and GigabitEthernet 1/0/2 is in the forwarding state (marked by a continuous line).

2.1.2  Master Port

The master port of a smart link group is a port role specified using commands. It can be an Ethernet port (electrical or optical), or an aggregate interface.

As shown in Figure 2 , the active port in the smart link group configured on Switch D is the master port GigabitEthernet 1/0/1, while that in the smart link group on Switch E is the slave port GigabitEthernet 1/0/2. Although GigabitEthernet 1/0/1 of Switch E is blocked, it is still the master port.

2.1.3  Slave Port

The slave port of a smart link group is another port role specified using commands. It can be an Ethernet port (electrical or optical), or an aggregate interface. The link on which the slave port resides is called the backup link.

As shown in Figure 2 , the blocked port in the smart link group on Switch D is the slave port GigabitEthernet 1/0/2, while that in the smart link group on Switch E is the master port GigabitEthernet 1/0/2. Although GigabitEthernet 1/0/1 of Switch E is in the forwarding state, it is still the slave port.

2.1.4  Protected VLAN

A protected VLAN is a VLAN that carries user data traffic within a smart link group. A port can belong to multiple smart link groups, each having different protected VLANs. Each smart link group independently calculates the state of its own member ports.

As shown in Figure 2 , Smart Link Group 1 and Smart Link Group 2 are created on Switch D. Smart Link Group 1 protects VLANs 1 to 10, while Smart Link Group 2 protects VLANs 11 to 20. In this way, the traffic of the two VLAN groups can be uplinked through different ports.

2.1.5  Control VLAN

1. Transmit control VLAN

A transmit control VLAN is a VLAN used by a smart link group to broadcast flush messages. For more information about flush messages, refer to Flush Message.

As shown in Figure 2 , if Switch D and Switch E are enabled to send flush messages, when link switchover occurs, the devices broadcast flush messages in the transmit control VLANs using the new primary links.

2. Receive control VLAN

A receive control VLAN is a VLAN used by an upstream device to receive and process flush messages.

As shown in Figure 2 , if the upstream devices (Switch A, Switch B, and Switch C) of Switch D and Switch E can identify flush messages, and are enabled to receive and process flush messages, when link switchover occurs, the upstream devices process flush messages from the receive control VLANs, and refresh their MAC address forwarding entries and ARP entries.

2.1.6  Flush Message

When link switchover occurs in a smart link group, the old forwarding entries are no longer useful for the new topology. Therefore, all devices in the network need to refresh their MAC address forwarding entries and ARP entries. Smart Link notifies devices to refresh their MAC address forwarding entries and ARP entries by sending flush messages to them.

A flush message uses the IEEE802.3 encapsulation format. It contains fields such as Destination MAC Address, Source MAC Address, Control VLAN ID, and VLAN Bitmap. Figure 3  shows the format of a flush message.

Figure 3  Flush message format

l              The Destination MAC Address field indicates an unknown multicast address. 0x010F-E200-0004 is the destination MAC address of a flush message.

l              The Source MAC Address field indicates the bridge MAC address of the device that sends the flush message.

l              Currently, only one control type, 0x01, is available, which is used to delete MAC address forwarding entries and ARP entries.

l              The current control version is 0x00, which is used for the expansion of later versions.

l              Device ID indicates the bridge MAC address of the device that sends the flush message.

l              Control VLAN ID indicates the ID of the transmit control VLAN.

l              Auth-mode indicates the authentication mode. It is used in conjunction with the Password field for subsequent security expansion.

l              VLAN bitmap is used to carry a list of VLANs whose MAC address table entries need to be refreshed.

l              FCS indicates the frame check sequence, which is used to check bit errors in frames.

2.2  Smart Link Mechanism

This section uses the network shown in Figure 4  to describe the smart link mechanism as the link status transiting from normal, to faulty, and then to recovery.

Figure 4  Smart Link mechanism

2.2.1  Link-Normal Operating

GigabitEthernet 1/0/1 and GigabitEthernet 1/0/2 of Switch A form a smart link group, with the former as the master port and the latter as the slave port. When both uplinks are healthy, the master port is in the forwarding state, while the slave port is in the standby state, and the links on which the two ports are seated respectively are called the primary link and the backup link. In this case, data is transmitted along the link indicated by the blue line. There is no loop in the network, hence no broadcast storms either.

2.2.2  Link-Faulty Handling

When the primary link on Switch A fails, the master port GigabitEthernet 1/0/1 transits to the standby state, while the slave port GigabitEthernet 1/0/2 transits to the forwarding state. A link switchover occurs. After the link switchover, the MAC address forwarding entries and ARP entries kept on the devices in the network may become incorrect, and need to be refreshed, so that traffic can be rapidly switched to another link, thus avoiding traffic loss. Currently, two mechanisms are available for refreshing MAC address forwarding entries and ARP entries:

1. Flush message-notified refreshing

This mechanism is applicable when the upstream devices (such as Switch B, Switch C, and Switch D in Figure 4 ) support Smart Link and are able to recognize flush messages.

To enable rapid link switchover, you need to enable Switch A to send flush messages, and all upstream devices’ ports that are on the dual uplink network to receive and process flush messages.

(1)        After link switchover occurs on Switch A, flush messages are sent along the new primary link, that is, through GigabitEthernet 1/0/2. The VLAN Bitmap field is filled by the protected VLAN IDs of the smart link group where GigabitEthernet 1/0/1 is in the forwarding state before the link switchover, and the Control VLAN ID field is filled by the transmit control VLAN ID of the smart link group.

(2)        When an upstream device receives a flush message, it determines whether the transmit VLAN ID carried in the flush message is among the receive control VLAN list configured on the receiving port. If not, the upstream device simply forwards the flush message. If yes, the device retrieves the VLAN Bitmap information from the flush message, and deletes the MAC address forwarding entries and ARP entries learned in the VLANs specified in the VLAN Bitmap field.

After that, when Switch D receives a data packet destined for Switch A: if Layer 2 forwarding is required, it broadcasts the packet at Layer 2; if Layer 3 forwarding is required, it updates the ARP entries using the ARP probing mechanism, and then forwards the packet. In this way, data traffic can be forwarded correctly.

 

&  Note:

l      To ensure that flush messages can be correctly transmitted within the transmit control VLAN, make sure that all ports on the dual uplink network belong to the transmit control VLAN. Otherwise, flush messages cannot be successfully transmitted or forwarded.

l      You are recommended to send flush messages tagged. To send flush messages untagged, you need to ensure that the default VLAN of the peer port is the same as the transmit control VLAN. Otherwise, flush messages will not be transmitted within the transmit control VLAN.

 

2. Uplink traffic-triggered refreshing

This mechanism is applicable when devices that do not support Smart Link or other vendors’ devices are connected. In such cases, the refreshing of the MAC address forwarding entries and ARP entries is triggered by uplink traffic.

l              If no uplink traffic from Switch A arrives at Switch D to trigger the refreshing of the MAC address forwarding entries and ARP entries, when Switch D receives a data packet destined for Switch A, it still forwards the packet out GigabitEthernet 1/0/1, and thus fails to send the packet to Switch A. The MAC address forwarding entries and ARP entries will not be refreshed until their aging timers expire.

l              Because the MAC address forwarding entries and ARP entries on Switch A are also incorrect, Switch A cannot send uplink traffic until its MAC address forwarding entries and ARP entries age out and new MAC address forwarding entries and ARP entries are learned. When uplink traffic from Switch A arrives at Switch D through GigabitEthernet 1/0/2, Switch D refreshes its own MAC address forwarding entries and ARP entries. After that, when Switch D receives a data packet destined for Switch A, it forwards the packet out GigabitEthernet 1/0/2. This time the packet can reach Switch A through Switch C.

When the first mechanism is used, the upstream devices refresh their MAC address forwarding entries and ARP entries based on received flush messages, instead of waiting until these entries age out. In this way, the time required for refreshing MAC address forwarding entries and ARP entries is considerably reduced. Normally, a link switchover is completed within a few subseconds, and practically no traffic is lost.

2.2.3  Link-Recovery Working Modes

Smart Link supports two working modes: role preemption and non-role preemption. Under different modes, the port state changes are different:

l              If role preemption is configured, when the primary link recovers, the master port enters the forwarding state and takes over the traffic, while the slave port enters the standby state. The slave port transits from standby to forwarding only when the primary link fails.

l              If non-role preemption is configured, when the primary link recovers, the slave port remains in the forwarding state, while the master port remains in the standby state, so as to keep the traffic stable.

As shown in Figure 4 , if role preemption is configured on the smart link group on Switch A, when the link of GigabitEthernet 1/0/1 on Switch A recovers, GigabitEthernet 1/0/2 is immediately blocked and transits to the standby state, while GigabitEthernet 1/0/1 transits to the forwarding state. If non-role preemption is configured, when the link of GigabitEthernet 1/0/1 on Switch A recovers, GigabitEthernet 1/0/1 remains in the standby state, and no link switchover occurs, thus keeping the traffic stable.

2.3  Load Sharing Achieved Through Smart Link

Data traffic of multiple VLANs may exist in a dual uplink network. By configuring Smart Link, you can achieve load sharing in the network, that is, transmit traffic of different VLANs along different paths. By configuring the two ports on the dual uplinks as members of two smart link groups (each having its own protected VLANs), each port having the opposite roles in different smart link groups, you can enable protected VLAN traffic of different smart link groups to be transmitted along different paths, thus achieving load sharing in the network.

You are recommended to configure role preemption in the smart link groups. Otherwise, load sharing will be interrupted upon the first link switchover. The reason is as follows: when non-role preemption is configured in the smart link groups, traffic is shared on the two uplinks in the beginning. When either uplink fails, all traffic is transmitted along one path, and remains that way even after the failed uplink recovers. Therefore, load sharing cannot last in the network.

As shown in Figure 4 , two smart link groups are created on Switch A and have different protected VLANs. Both of them are configured with the role preemption mode. Smart Link Group 1’s master port is GigabitEthernet 1/0/1, slave port is GigabitEthernet 1/0/2, and protected VLANs are VLAN 1 through VLAN 10; Smart Link Group 2’s master port is GigabitEthernet 1/0/2, slave port is GigabitEthernet 1/0/1, and protected VLANs are VLAN 11 through VLAN 20. Both smart link groups’ master ports are in the forwarding state. Thus configured, traffic of VLAN 1 through VLAN 10 is transmitted along the link indicated by the blue line, and that of VLAN 11 through VLAN 20 is transmitted along the link indicated by the red line. In this way, VLAN traffic is shared on the two uplinks.

2.4  Restrictions

l              You cannot assign an STP- or RRPP-enabled port to a smart link group as a member port.

l              You cannot assign a member port of an aggregation group or service loopback group to a smart link group as a member port.

Monitor Link Overview

3.1  Background

Figure 5  Monitor Link background

As shown in Figure 5 , a smart link group is configured on Switch A for link redundancy purpose, with GigabitEthernet 1/0/1 as the master port, and GigabitEthernet 1/0/2 as the slave port.

When the primary link on which GigabitEthernet 1/0/1 resides fails, traffic on it switches to the backup link on which GigabitEthernet 1/0/2 resides within a few subseconds. Smart Link delivers reliable link redundancy and rapid convergence.

However, when the link on which the uplink port GigabitEthernet 1/0/1 of Switch B resides fails, link switchover will not happen in the smart link group configured on Switch A because the link on which the master port GigabitEthernet 1/0/1 resides is healthy. But in fact, traffic of Switch A can no longer reach Switch D through GigabitEthernet 1/0/1, and the traffic is thus interrupted. To address this problem, the Monitor Link technology is introduced.

3.2  Benefits

Monitor Link is developed to complement the Smart Link feature. By monitoring the uplink, and synchronizing the downlink with the uplink, Monitor Link triggers the switchover between the primary and backup links in a smart link group, thus perfecting the link redundancy mechanism of Smart Link.

Monitor Link Implementation

4.1  Basic Concepts in Monitor Link

4.1.1  Monitor Link Group

A monitor link group is a set of uplink and downlink ports. Downlink ports adapt to the state changes of uplink ports.

Figure 6  Monitor Link basic concepts

As shown in Figure 6 , ports GigabitEthernet 1/0/1, GigabitEthernet 1/0/2, and GigabitEthernet 1/0/3 of Switch A form a monitor link group.

4.1.2  Uplink Port

An uplink port is a monitored port in a monitor link group. It is a port role specified using commands. It can be an Ethernet port (electrical or optical), or an aggregate interface.

As shown in Figure 6 , GigabitEthernet 1/0/1 of Switch A is the only uplink port of the monitor link group configured on the device.

For a monitor link group that has multiple uplink ports, as long as at least one of its uplink ports is in the forwarding state, the monitor link group is up. However, when all uplink ports of the monitor link group fail, the monitor link group goes down, shutting down all the downlink ports. If no uplink port is specified in a monitor link group, the system considers the monitor link group’s uplink ports to be faulty, and thus shuts down all the downlink ports in the monitor link group.

4.1.3  Downlink Port

A downlink port is a monitoring port in a monitor link group. It is another port role specified using commands. It can be an Ethernet port (electrical or optical), or an aggregate interface.

As shown in Figure 6 , GigabitEthernet 1/0/2 and GigabitEthernet 1/0/3 of Switch A are two downlink ports of the monitor link group configured on the device.

 

&  Note:

When a monitor link group’s uplink ports recover, only downlink ports that were blocked due to uplink port failure will be brought up. Downlink ports manually shut down will not be brought up automatically. The failure of a downlink port does not affect the uplink ports or other downlink ports.

 

4.2  Monitor Link Mechanism

As shown in Figure 7 , to provide reliable access to the Internet for the hosts, a smart link group is configured on Switch A. GigabitEthernet 1/0/1 is the master port of the smart link group, and is in the forwarding state. GigabitEthernet 1/0/2 is the slave port.

Figure 7  Monitor Link mechanism

To avoid traffic interruption due to the failure of the link on which GigabitEthernet 1/0/1 of Switch B resides, configure a monitor link group on Switch B, and specify GigabitEthernet 1/0/1 as the uplink port, and GigabitEthernet 1/0/2 as the downlink port.

When the link on which GigabitEthernet 1/0/1 of Switch B resides fails, the monitor link group shuts down its downlink port GigabitEthernet 1/0/2, triggering a link switchover in the smart link group configured on Switch A.

When the link on which GigabitEthernet 1/0/1 of Switch B resides recovers, the downlink port GigabitEthernet 1/0/2 is also brought up, triggering another link switchover in the smart link group if role preemption is configured in the smart link group on Switch A.

Collaboratively, Monitor Link and Smart Link deliver reliable link redundancy and fast convergence for dual-uplink networks.

4.3  Restrictions

You cannot assign a member port of an aggregation group or service loopback group to a monitor link group as a member port.

Application Scenarios

5.1  Smart Link Combined with Monitor Link

Figure 8  shows a typical dual uplink network, which is a major application scenario for Smart Link and Monitor Link.

Figure 8  Collaboration between Smart Link and Monitor Link

In the above application environment, Smart Link is configured on Switch D and Switch E. By configuring multiple smart link groups on the devices and assigning different protected VLANs to them, you can transmit traffic of protected VLANs that belong to different smart link groups along different paths, thus achieving load sharing. When the link between Switch B and Switch D, or that between Switch C and Switch E fails, the smart link groups can immediately sense the link failure and switch over links.

To enable Switch D (or Switch E) to directly sense the failure of the link between Switch A and Switch B (or Switch C), you need to configure a monitor link group on Switch B (or Switch C), with GigabitEthernet 1/0/1 as the uplink port, and GigabitEthernet 1/0/2 and GigabitEthernet 1/0/3 as the downlink ports.

Upon detecting the failure of the link on which the uplink port resides, the monitor link group shuts down its downlink ports, triggering a link switchover in the smart link group configured on Switch D or Switch E. When the uplink port or link failure recovers, the downlink ports are brought up automatically, which enables Switch D (or Switch E) to sense the state changes of the link between Switch A and Switch B (or Switch C).

5.2  Smart Link Cascaded with Monitor Link

Figure 9  shows an application scenario where Smart Link is cascaded with Monitor Link to achieve more reliable link redundancy.

A smart link group member port can be assigned to a monitor link group as its uplink member port. By using the Smart Link and Monitor Link technologies together, you can cascade the backup links.

To do that, you need to configure the two member ports of a smart link group as the uplink ports of a monitor link group, with the peer of the monitor link group’s downlink port as the master or slave port of another smart link group, as shown in Figure 9 .

Figure 9  Cascaded smart link groups

Table 1  lists the roles of ports on Switch C, Switch J, and Switch F in Smart Link and Monitor Link shown in Figure 9 .

Table 1  Smart Link and Monitor Link port roles

Switch

Smart Link Group 1

Monitor Link Group 1

Master Port

Slave Port

Uplink Port

Downlink Port

Switch C

GE1/0/1

GE1/0/2

GE1/0/1, GE1/0/2

GE1/0/3

Switch J

GE1/0/1

GE1/0/2

GE1/0/1, GE1/0/2

GE1/0/3

Switch F

GE1/0/1

GE1/0/2

No monitor link group is configured.

 

As shown in Figure 9 , the red lines indicate the first level Smart Link backup uplinks, and the blue lines indicate the second level Smart Link backup uplinks.

5.3  Smart Link Combined with RRPP

Figure 10  shows a hybrid network of Smart Link and RRPP.

In this network, RRPP is enabled on Switch A, Switch B, Switch C, and Switch D to provide link redundancy.

If STP is used to provide link redundancy, you need to enable STP on all ports connecting Switch C, Switch D, and Switch E. Because RRPP, which is mutually exclusive with STP, is enabled on the two ports connecting Switch C and Switch D, you can achieve link redundancy for Switch E by configuring a smart link group on Switch E, which is not only simpler than configuring RRPP subrings on Switch C, Switch D, and Switch E, but also applies to the scenario where Switch E does not support RRPP.

Figure 10  Smart Link-RRPP hybrid network

 

 

 

 

Copyright ©2008 Hangzhou H3C Technologies Co., Ltd. All Rights Reserved

Extracting and copying partial or whole contents of the document without H3C’s written permission is prohibited.

The information is this document is subject to due modification.