In this paper, we propose a new protection scheme, which we term partial path protection (PPP), to select endto-end backup paths using local information about network failures. PPP designates a different restoration path for every link failure on each primary path. PPP also allows reuse of operational segments of the original primary path in the protection path. A novel approach used in this paper is that of a dynamic call-by-call model with blocking probability as the performance metric, this model is in contrast with traditional capacity-efficiency measurement for batch call arrivals. Additionally, we show that a simple method based on shortest path routing for which primary paths are selected first is more effective than a greedy approach that minimizes, for each call arrival, the number of wavelengths used by the primary and backup path jointly.