dll stablity
There is a difference w.r.t. stabiltiy.
A PLL will at least have one pole (ideal integrator) from the VCO (voltage to frequency conversion, so you have to integrate to get the phase transfer -> VCO transfer = Kvco/p with Kvco in rad/V*s).
To get zero phase difference you'll have to create one extra integrator in the loop, so you end up with a second order system, which can get instable if you don't introduce a zero.
The DLL does not have the pole from the VCO: the delay line is a voltage to phase conversion (delay line gain rad/V)
So when introducing one integrator in the loop to get zero phase difference, you'll end up with a first order system which is unconditionally stable.
Still you have to watch out for parasitic poles, introduced inside your loop, so you might end up with a second order system.
p.s. a DLL can give better noise performance at low frequency offsets w.r.t. a PLL, because the regenerative nature of the VCO will increase the noise near the carrier. A DLL will be "reset" by the reference on each input edge.
Sceptre