As many in this forum know, I'm of a different opinion, because I think it's very clear that there is no instantaneous collapse, because the most comprehensive quantum theory we have is relativistic quantum field theory, and it is from the very beginning constructed in a way that there are no faster-than-light causal connections between events possible.
If you have an entangled pair of particles (or photons, which are most easy to operate with in the lab when it comes to entanglement) and you measure, e.g., the momentum of one of the particles, nothing happens to the other particle, at least not for the time it takes to send a signal from you to the other particle, i.e., you need at least the time light needs to travel from your place to the other particle's place.
That you know immediately the value of the other particle's momentum is due to the preparation of the two particles in the entangled state, i.e., although the single-particle momenta of both particles are indetermined before the measurement, they are strongly correlated due to the preparation in the entangled state in the very beginning (e.g., by decay of a heavier particle at rest into two ligher particles, which then have strictly opposite momentum because of momentum conservation, but you don't know in which direction both fly), i.e., you know when measuring the momentum that this particle now has momentum ##\vec{p}##, then you know that somebody measuring the other particle's momentum must bei ##-\vec{p}##. It doesn't matter, who measures his particle's momentum first or if you do the measurement at the same time (in your common rest frames), you'll always find a random distribution of momenta (or directions of momenta in our decay example) but you find 100% correlation, i.e., the outcomes of the particle-momenta measurements always give opposite momenta.
I'm a proponent of the minimal statistical interpretation, i.e., there is nothing to be known about particles than the probabilities for the outcome of measurements, quantum (field) theory provides. What happens to the particles in the process of measurement depends of course on how specifically your measurement apparatus works, i.e., whether or not after the measurement you have a particle with the measured value of the observable you measured, depends on the measurement device. For sure, there's no instantaneous influence of far distant local measurements on two particles, no matter whether they were prepared in an entangled or an unentangled state.
I'm also agnostic about the question, how it comes that we always measure (within the accuracy of the measurement device) one specific value for an observable that was undetermined by the preparation of the measured quantum system. It's just the construction of the apparatus, and all we know before the measurement is the probability for the outcome of this measurement. At the present state of our knowledge that's really all we can know about quantum systems.
The apparent classical behavior of macroscopic objects, including measurement devices, is well understood by quantum-statistical physics, i.e., considering only the relevant macroscopic observables (e.g., the position of the center of momentum of a planet moving around the Sun), are spatio-temporal averages over many microscopic degrees of freedom, and those tend to behave in almost all cases classical.