The cosmic microwave background is anisotropic: the temperature is distributed as a dipole with the temperature at the poles differing by ±0.00335 K from the mean. This defines a Lorentz frame: we can boost by several hundred km/s to make the dipole vanish, on average. This new frame could be considered the rest frame of the CMB. The laws of Physics are Lorentz invariant, so this appears to require spontaneous breaking of Lorentz symmetry by the CMB. My question is: how and when is this breaking supposed to have occurred? Must the frame have been chosen at the start of the Universe, or is it possible for a Lorentz-invariant early Universe to break its symmetry at recombination? Could there be topological defects that are remnants of recombination when we try to define the CMB frame globally?