Due to their inherent parallel and non-deterministic nature, P system implementations require vast computing and storage resources. This significantly limits their applications, even more so when the calculation of all possible evolutions of the P system is required. This article exposes the scalability possibilities available with the Big Data ecosystem for P systems implementations, using Map Reduce parallelism to build the P system computation tree. The Hadoop based implementation is then used for generating test suites for cell like P systems, in particular for context-dependent rule coverage testing. Our preliminary evaluations on a benchmark of automatically generated P systems confirms that the proposed approach scales well.