For any FTL architecture, the frequency of garbage collection activities depends on host usage model: a defrag operation can be never required if the host writes sequentially all logical addresses while it can be required at each host access for random writes.