| This section discusses MPLS fault management requirements that serve as input to developing mechanisms to manage MPLS-based networks. It also covers key attributes required to manage and operate IP/MPLS-based networks within the framework of the FCAPS management model. Maintaining core integrity is key in identifying requirements for MPLS OAM. The primary objective is to reduce cost by minimizing service interruptions. Minimal requirements include the ability to detect and diagnose a break in the LSP data path and identify the source of the failure. MPLS OAM OAM solutions provide the following capabilities: Detection, diagnosis, and localization of broken label-switched path (LSP) Any OAM solution should provide the capability to diagnose and detect a broken LSP because diagnosing a broken LSP and isolating the failed resource in the path are required. This is true for misbranching defects, which are particularly difficult to specify recovery. The fundamental requirement, therefore, is to detect and diagnose an MPLS LSP. Additionally, the path trace function must have the ability to support equal cost multipath (ECMP) scenarios. ECMP is often used for both load sharing and redundant path capabilities. The OAM mechanism should support equal cost multipath LSPs ECMP scenarios appear when several LSPs can be used to carry data from the head end to the tail end. In this particular situation, the OAM mechanism should be able to exercise and verify all those paths that might transport data within a reasonable amount of time. Unfortunately, there is no standard for the load-sharing algorithm, but any function needs to be capable of detecting failures on all operational paths. This is because a failure of any branch can lead to loss of traffic, regardless of the load-sharing algorithm. The ability to raise an alarm when failures are detected A defect event in a lower layer should not cause multiple alarms to be raised. This capability is required for alarm suppression and root cause analysis of a fault condition. Upon detection of a broken LSP, the correct alarm/notification should be sent to the LSRs or the network management system. For the sake of example, if the LSP is to carry Layer 2 circuits, a defect at the LSP level should not target multiple alarms at the Layer 2 level. Those mechanisms are required to measure different aspects of SLAs, such as jitter, latency, and packet loss.
One extra parameter of interest for service providers is the network availability. (The definition of network availability likely differs from one provider to the other. At a high-level overview, you can define it as a function of jitter, packet loss, and latency.) |
