Classification of life distributions based on important properties like failure rate and its averages, residual life and its summary measures, properties of the distribution or density function, coordinate sub-tangent of the survival curve, etc. is required for a non-parametric or semi-parametric approach in several problems in reliability and survival analysis. Thus, to find bounds on reliability and to work out optimal maintenance policies, a realistic assumption likely to hold good in many situations would be that the underlying failure time distribution has an increasing failure rate.