The elastic net clustering algorithm (ENCA) can typically provide an effective way for classifying non-linearly separable data. However, the computation time it takes will be significantly increased for large datasets. To deal with this issue, a parallel version of the ENCA, built on the Apache Spark framework, called parallel elastic net clustering algorithm (PENCA), is presented in this paper. To evaluate the performance of the proposed algorithm, it is compared with ENCA and two well-known clustering algorithms, k-means and genetic k-means algorithm (GKA). The results show that PENCA not only outperforms k-means and GKA in terms of the accuracy rate, it also provides an efficient way to reduce the response time of ENCA-based clustering algorithms for large-scale datasets.