AAQIB IQBAL is talking about "the derivative" as if it were a single number and, for functions of several variables, that is not true.
Simon Bridge is talking about the partial derivatives- imagine walking along a hillside (representing the surface z= f(x,y)) moving due east (the positive x-axis). The rate at which you go up or down is the partial derivative with respect to x. Moving northward (the positive y-axis) the rate at which you go up or down is the partial derivative with respect to y.
The closest thing to "THE" derivative is the gradient, [itex]\nabla f[/itex], which is a vector pointing in the direction of fastest increase whose length is the rate of increase in that direction. Just as the "tangent line", having slope [itex]f'(x_0)[/itex], to a curve in the plane gives the "best" linear approximation to y= f(x) close to [itex]x= x_0[/itex], the "tangent plane, [itex]z= f_x(x_0,y_0)(x- x_0)+ f_y(x_0,y_0)(y- y_0)+ z_0[/itex] gives the best linear approximation to z= f(x,y) close to [itex](x_0, y_0)[/itex]. In terms of the gradient, that can be written [itex]z= \nabla f(x_0,y_0)\cdot <x- x_0, y- y_0>+ z_0[/itex] where that multiplication is the dot product of vectors.
The formula your book gives is actually the equation of that tangent plane where the left side is "[itex]z- z_0[/itex]", [itex]\Delta x= x- x_0[/itex], [itex]\Delta y= y- y_0[/itex] and the last two terms are the "error" terms- the distance from the tangent plane to the actual surface.