Abstract: A first set of garbage collection (GC) features and non-GC features associated with a storage system are received, the first set of features being associated with a predetermined start date and a time window. A learning equation is generated having a plurality of vectors of GC features and a plurality of vectors of non-GC features. For a current iteration representing a current GC process, it is determined whether a first prior GC process was started within the time window. An entry of vectors of the non-GC features of the learning equation is populated based on corresponding feature values of the first set of non-GC features, in response to determining that the first prior GC process was started within the time window. A predetermined regression algorithm is applied to the learning equation to generate a GC duration predictive model to predict a GC duration of a subsequent GC process.

Abstract: Mechanisms for predicting a GC duration are described herein. In one embodiment, the mechanisms include receiving a first set of features determined based on current operating status and prior garbage collection (GC) statistics of a first storage system. In one embodiment, the mechanisms include predicting a GC duration of a first GC process being performed at the first storage system by applying a predictive model on the first set of features, wherein the predictive model was generated based on a second set of features received periodically from a plurality of storage systems.