In IP networks Backup paths are widely used to protect IP links from failures. Still, existing solutions such as the commonly used independent model and Shared Risk Link Group (SRLG) model will not exactly redirect the correlation between IP link failures, and which may not choose reliable backup paths. We propose a cross-layer approach for minimizing routing disruption caused by IP link failures. We develop a probabilistically correlated failure (PCF) model to quantify the impact of IP link failure on the reliability of backup paths. With this model, we propose an algorithm to pick several reliable backup paths to defend each IP link. If an IP link fails, its traffic is divided into several backup paths to ensure that the rerouted traffic load on each IP link does not exceed the usable bandwidth. Experimental results show that two backup paths are adequate for protecting a logical link. Compared with existing works, the backup paths selected by our approach are at least 18 percent more reliable and the routing disruption is reduced by at least 22 percent. Unlike prior works, the proposed approach prevents the rerouted traffic from interfering with normal traffic.