The 802.1Qbe—Multiple I-SID Registration Protocol (MIRP) standard provides the ability to flush learned MAC address registration entries held in the filtering database of an I-component on a per I-SID basis. The backbone service instance identifier (I-SID) is a field in the backbone service instance tag which identifies the backbone service instance of a frame. MIRP defines mechanisms for I-SID flushing, and has the required capabilities to handle topology changes that occur in networks attached to a provider backbone bridged network. A backbone edge bridge (BEB) signals to other potentially affected BEBs, the need to alter certain learned associations between customer MAC addresses and backbone MAC addresses. In the absence of MIRP, customer connections across a provider backbone network can take several minutes to restore connectivity after a topology change in an access network.
In prior releases, PBB traffic was dropped for a MAC aging cycle when bridge forwarding topology changes occurred (due to unavailable ports or spanning tree topology changes) in a PBB edge bridge domain. This resulted in severe limitations for the use of PBB bridges.
Cisco ASR 9000 Series Aggregation Services Routers now support a simplified implementation of the MIRP protocol known as the Multiple I-SID Registration Protocol Lite (MIRP-Lite). The MIRP-Lite feature enables detection of a topology change at a site. A specially defined packet is flooded to all remote edge sites of the PBB network when a site detects a topology change. At the sender site, I-SID of the I-component is placed in the I-TAG of the frame header to specify the I-SID that needs a MAC flush. At the receiver site, each PBB edge switch performs I-SID checking. If the I-SID matches one of the I-components, the MAC in the I-component is flushed.
The use of MIRP in 802.1ah networks is illustrated in the following figure.
Figure 4. MIRP in 802.1ah Networks
Device DHD1 is dual-homed to two 802.1ah backbone edge bridges (BEB1 and BEB2). Assume that initially the primary path is through BEB1. In this configuration BEB3 learns that the host behind DHD1 (with MAC address CM1) is reachable via the destination B-MAC M1. If the link between DHD1 and BEB1 fails and the host behind DHD1 remains inactive, the MAC cache tables on BEB3 still refer to the BEB1 MAC address even though the new path is now via BEB2 with B-MAC address M2. Any bridged traffic destined from the host behind DHD2 to the host behind DHD1 is wrongly encapsulated with B-MAC M1 and sent over the MAC tunnel to BEB1, where the traffic drops.
To circumvent the dropping of traffic when the link between DHD1 and BEB1 fails, BEB2 performs two tasks:
- Flushes it’s own MAC address table for the service or services.
- Requests the remote PE that receives the MIRP packet to clear it’s own MAC table. The MIRP message is transparent to the backbone core bridges (BCBs). The MIRP message is processed on a BEB because only BCBs learn and forward, based on B-MAC addresses and they are transparent to C-MAC addresses.
MIRP triggers C-MAC address flushing for both native 802.1ah and PBB over VPLS.
The following figure shows the operation of the MIRP.