Is there any good alternative technique to k-means for clustering data? Should we use SAS, R, or other?

I would say use what you are mastering and available for you, K-means or any statistical concepts and methods could be used with different tools and coding languages.

K means is a statistical concept to estimate the variance from the mean on a range of variables. The alternative is to evaluate the exact deviation on a line by line basis, which can be resource intensive on large data sets.

Joan, there are other alternatives to K-Means: DBSCAN, K-Harmonic, Fuzzy K-Means.., but sincerely I am not used to those.

Joan, you may use an open source tool called Gephi (https://gephi.org/users/download/) for clustering. It provides different algorithms and methods for clustering and can deal with large amounts of data.  Cheers