Some links for you :

• Software : Try Deep Sky Stacker. It's freeware and designed for astrography. The site has some useful tips on technique and method as well, so read those even if you don't use the software.

• General purpose software :

  • I do use Darktable for general photography processing, but it's not designed for astrography. For very tough RAW files I'm trying to squeeze the most out of I sometimes use RawTherapee which is slightly more specialized and has more control for demosaicing algorithm choices, but Darktable is now very good. Experiment with the Noise Reduction options on Darktable as you can do channel and range specific adjustments with masks that will make this easier.
  • I use GIMP. If you're not already using the G'Mic plugin then get it. There are a lot of great processing options in there, including Fourier transform and a lot of general purpose noise reduction stuff. GIMP is fine for general image processing and finalizing images, but I would not use until you've done the "heavy lifting" in something like Deep Sky Stacker. I would definitely not use GIMP for stacking astrography images.
  • There's an older software package called Iris which is possibly worth checking out for astrophotography. It might be a useful complement to Deep Sky Stacker.

Regarding general approach you need to do some experimentation to get the best from your particular mix of technique and equipment and software. Everyone has different kit and expectations and so they need different approaches.

I'm not sure you need dark frame subtraction per se on a modern camera like the 70D if you're also doing stacking and follow the very detailed suggestions in Deep Sky Stacker. Modern sensors have a lot of noise suppression for this kind of noise going on at a chip level and I'm not sure dark frame subtraction is quite as effective as it once was "in ye old days". In any case, have a good read of the material on Deep Sky Stacker's site and form your own view.

There's a technique which is sometimes useful nowadays which means you shoot at low ISO (whatever is optimal for dynamic range for your camera) and raise ISO in post processing. This is sometimes called ISOless shooting. The idea with modern sensors is that some of them produce less noise doing this than raising ISO in-camera and you preserve a larger dynamic range. This does not work on all cameras and can be done by the camera "under the hood" anyway. I would not worry about this unless you have run out of other options.