Xrrp - Router Redundancy Protocol - HP 5300 Reviewer's Manual

•
Unlike a router, no packet modification is required as it travels through the switch.
Other HP ProCurve switches that support Switch Meshing, such as the HP ProCurve Switch 4000M
family, will work together in a mesh with the HP ProCurve Switch 5300xl Series. There are a few
corner case caveats in this type of mixed environment that are covered in the HP ProCurve 5300xl
documentation. The HP ProCurve 5300xl documentation can be found at:
http://www.hp.com/go/hpprocurve
A white paper with more details on Switch Meshing can be found in the information library on HP's
networking web site at http://www.hp.com/go/hpprocurve.
2.3.4 XRRP – Router Redundancy Protocol
One form of high availability in a Layer 3 environment is having two routing switches back each other
up. In the event of a connection failure with one of the routing switches, the other routing switch
transparently takes over the routing function. XRRP, the XL Router Redundancy Protocol, provides the
mechanism in the HP ProCurve Switch 5300xl Series routing switches for this backup functionality.
Similar in concept to VRRP (Virtual Router Redundancy Protocol), XRRP presents a virtual router to
the end node connections whose IP and MAC address can transition from the master HP ProCurve
5300 to the backup HP ProCurve 5300 on master 5300 interface failure. Since the end node connections
are tied to the virtual router using the virtual router IP and MAC addresses, they are unaware as to
whether the actual physical routing services are being provided by the master 5300, or, after a switch-
over, to the backup 5300, making any switch-over transparent to the end nodes. An XRRP interface
failure is defined as the inability of the master physical interface in the 5300 pair to be heard by the
backup interface. This could be caused by a cable failure, module failure, whole 5300 failure, or
operator error (such as a disconnected cable).
Some XRRP specifications:
•
Number of physical routers in a backup group (XRRP calls this a 'protection domain'): 2
•
Number of protection domains allowed per VLAN: 16
•
Time to failure detection and switchover: default – 15 seconds. Minimum time by making a
configuration change – 3 seconds. If a VLAN is lost on one of the 5300 pairs, but the 5300
doesn't go down, fail-over occurs in under 1 second as the 5300 with the failed VLAN
reports the loss directly to the other 5300 via a different VLAN.
•
Backed-up interfaces should be configured identically between the routers. XRRP checks
and warns if interface configurations do not match.
•
Master interfaces can be split between the two 5300 switches, allowing a split of traffic
between the two 5300s under normal network conditions.
•
If a failure is detected on any master interface, all the XRRP master interfaces on that 5300
are switched over to the back-up router. This allows easier troubleshooting, or module or
box replacement. When all interfaces on the failed 5300 are restored, the master
relationship is re-established as it was before the fail-over. There is a time interval (XRRP
fail back – default 10 seconds, configurable to 999 seconds) before master re-establishment
can take place to prevent master interface flapping due to interfaces that may be going up
and down.
•
Those interfaces not defined as part of the XRRP set on a 5300 will continue to run on that
5300 (unless, of course, the whole 5300 is down) even as the XRRP interfaces switch over to
the backup 5300. This has value for low priority interfaces where the cost of redundant
resources across two 5300s for these interfaces is not cost justified.
© Hewlett-Packard Co. 2002, 2003
HP ProCurve Switch 5300xl Series Reviewer's Guide
under the Technical Support section.
Rev 1.1 – 2/11/2003
http://www.hp.com/go/hpprocurve
Page 16 of 35