Glacier and ice-sheet motion is fundamental to glaciology. However, there is still no clear consensus for the optimal way to describe the sliding component of glacier and ice-sheet dynamics. Typically, sliding is parametrised using a traction coefficient linked to a given theory describing one or a limited set of sliding processes. However, this precludes the possibility of multiple overlapping, spatio-temporally varying sliding modes with inaccuracies resulting in model error propagation as the system evolves away from the conditions under which it was optimised for. Here, we describe glacier sliding as a scale- and setting-dependent 'inner flow' that arises from multiple overlapping sub-processes, including viscous ice deformation and obstacle resistance (or 'form drag'). The corresponding 'outer flow' then accounts for ice deformation that is minimally influenced by bed properties. Following this framework, we suggest that a rough-smooth bed division may be more consequential than a hard-soft one, and that the importance of 'Iken's bound' may be significantly reduced if viscous ice deformation becomes significant in a given region within the inner flow, which may explain the persistent functionality of power-law sliding in large-scale process-agnostic sliding studies over rough topography. Further, reviewing observation-based studies and considering the diversity of sliding processes and settings, we suggest that a simple 'unified' sliding relationship controlled by a single tunable coefficient may not be realistic. However, we show that given reasonable assumptions, sliding relationships should generally fall within a sum of regularised-Coulomb and power-law components, and suggest that a or compound, or power-law relationship with careful consideration of the power value may provide the most flexible way to account for the varied net effect of compound sliding sub-processes.