Understanding Uplink Failure Detection
Uplink failure detection allows Juniper Networks EX Series Ethernet Switches to detect link failure on uplink interfaces and to propagate the failure to the downlink interfaces so that servers connected to those downlink interfaces can switch over to secondary interfaces.
Uplink failure detection supports network adapter teaming and provides network redundancy. In network adapter teaming, all the network interface cards (NICs) on a server are configured in a primary or secondary relationship and share the same IP address. When the primary link goes down, the server transparently shifts the connection to the secondary link. With uplink failure detection, the switch monitors uplink interfaces for link failures. When it detects a failure, it disables the downlink interfaces. When the server detects disabled downlink interfaces, it switches over to the secondary link to help ensure balanced traffic flow on switches.
This topic describes:
Uplink Failure Detection Overview
Uplink failure detection allows switches to monitor uplink interfaces to spot link failures. When a switch detects a link failure, it automatically disables the downlink interfaces in that group. The server that is connected to the disabled downlink interfaces triggers a network-adapter failover to a secondary link to avoid any information drop.
Figure 1 illustrates a typical setup for uplink failure detection.
For uplink failure detection, you specify a group of uplink interfaces to be monitored and downlink interfaces to be brought down when an uplink fails. The downlink interfaces are bound to the uplink interfaces within the group. If all uplink interfaces in a group go down, then the switch brings down all downlink interfaces within that group. If any uplink interface returns to service, then the switch brings all downlink interfaces in that group back to service.
Routed VLAN interfaces (RVIs) cannot be configured as uplink interfaces to be monitored.
The switch can monitor both physical-interface links and logical-interface links for uplink failures, but you must put the two types of interfaces in separate groups.
To detect failure of logical interfaces, the server must run some high level protocol such as keepalives between the switch and the server.
Failure Detection Pair
Uplink failure detection requires that you create groups that contain uplink interfaces and downlink interfaces. Each group includes one of each of the following:
A link-to-monitor interface—The link-to-monitor interfaces specify the uplink interfaces the switch monitors. You can configure a maximum of 48 uplink interfaces as link-to-monitor in a group.
A link-to-disable interface—The link-to-disable interfaces specify the downlink interfaces the switch disables when the switch detects an uplink failure. You can configure a maximum of 48 downlink interfaces as link-to-disable in a group.
The link-to-disable interfaces are bound to the link-to-monitor interfaces within the group. When a link-to-monitor interface returns to service, the switch automatically enables all link-to-disable interfaces in the group.