You should simply make each dimension (or attribute, or column) have zero mean and unit variance.
Check this pdf to brings each dimension of the SVM into the same magnitude.
The main advantage of scaling is to avoid attributes in larger numeric ranges dominating those in smaller numeric ranges. Another advantage is to avoid numerical difficulties during the calculation.
Kernel values usually depend on the inner products of feature vectors, e.g. the linear kernel and the polynomial kernel, large attribute values might cause numerical problems. We recommend linearly scaling each attribute to the range [-1,+1] or [0,1].
Hope this answer helps you!
If you want to study Artificial Intelligence Course and also want to go through Deep Learning Tutorial, then you can watch this video: