Out of curiosity, were you doing a high number of reads at the same second? And if so, was it the same batch processing method you were using, or was it random reads?