Yup, Az is right, go around the brim with a razor to get it started, then apply a huge force... of patience and leave it to sort itself out. 
I've chipped out boro glass being impatient with an ABS print on hairspray!
Re warping at the periphery of the bed:
If your SD4 print bed is anything like my later-model SD3, the heater pad is not the whole size of the bed, it's approximately a 150mm square in the middle of the 200mm square bed. The aluminium bed is supposed to conduct the heat and even it up a bit, but especially as heat loss is greatest from the outer parts of the bed (bare edges and the bits where the silicone heater pad itself doesn't provide any insulation, just the fiberglass matting), the temperature tends to drop off as you get to the outside of the bed.
Also, if your bed is a little bowed up in the middle, the glass will contact well in the middle but there will be an air-gap underneath at the edges, so the conduction into the glass won't be as good - this also makes the temps at the outside of the bed rather less than the middle.
There's lots of discussion about improving the bed heating, the goal being perfect flatness and even surface temperature. However, there's only so much you can do, so in reality you have to accept that the outer edges of *any* bed are going to be troublesome. If you're doing multiple pieces, reduce the count and keep them closer to the middle of the bed, and accept it will take more print jobs to complete. If you're doing large pieces then you'll probably have to look at using PLA instead of ABS, consider a split/glue job, or pump up the ambient temp with active heating...
SD3. Mk2b + glass, heated enclosure, GT2 belts, direct drive y shaft, linear bearings, bowden-feed E3D v5 w/ 0.9° stepper
Smoothieboard via Octoprint on RPi
