A first provider edge (PE) determines whether a link between the first PE and a second PE is a high latency link, where the second PE is a next hop of the first PE. When determining an actual latency corresponding to the link, the first PE compares the actual latency with a first latency threshold. When the actual latency is greater than or equal to the first latency threshold, it indicates that the actual latency exceeds a latency upper limit corresponding to the link, the first PE determines that the link is a high latency link. In this way, it is not necessary to wait for network management personnel to determine whether the link is a high latency link. After determining the actual latency corresponding to the link, the first PE can determine in time whether the link is a high latency link.