> You are absolutely right in that the CMB becomes Doppler shifted when you have some relative velocity. But Earth is not in the CMB rest frame: when we observe the CMB from Earth, we observe a dipole component to the CMB caused by the Earth's motion orbiting the Sun, the Sun's orbit around the Milky Way, and any velocity the Milky Way as a whole has. I was at a talk about the new Planck results last week, and saw a plot in which you can clearly see the dipole component in the raw data. You need to correct for this motion before you can even remotely see the anisotropies that are the interesting science goals of Planck.
> There is a rest frame in which the CMB is closest to isotropic (no dipole component), and this rest frame is special but not 'absolute'. This frame is effectively the 'center-of-momentum' frame of the observable universe, in which we expect the total momentum to be zero. We know from classical mechanics that for any system of objects, we can construct such a frame, and that it sometimes has useful properties for solving certain types of problems. But there is nothing 'absolute' about this rest frame, the laws of physics operate entirely the same.
> And so this is fine, because ultimately what relativity requires is that the laws of physics operate the same in every rest frame, not that every rest frame looks the same. Because the CMB is itself physical (made of photons) and was emitted by matter, it is entirely natural that it should be affected by frame transformations, and should look different if you shift to a frame that is moving differently than the emitting medium.
The frequency of light changes depending on how fast you are moving relative to it. Move away from a light source and the light still approaches you from the same speed, but is lower frequency. Rest here is where the frequency becomes "uniform" (ish) regardless of direction.
The laws of physics don't change in different frames (Einstein's assumption), but that doesn't mean that all frames are equivalent in other respects.