Well in heavy-ion collisions a nanosecond is like eternity. The usual scale in this physics are femto-meters (Fermi) ##1 \; \mathrm{fm}=10^{-15} \; \mathrm{m}##. In ultrarelativistic heavy-ion collisions such as performed at the CERN SPS, LHC and BNL RHIC you collide two heavy nuclei like Pb or Au. Usually they collide with some finite impact parameter, i.e., you have a non-central collision. You can think of this in terms of the Glauber model, i.e., just by geometry, a part of the nucleons really collide (or at this energies the corresponding partons within these nucleons). They are called participants, and the rest are the spectators.
In this collision some very hot and dense lump of matter is created, and that's the purpose to investigate these collisions. The reason is that there is overwhelming evidence that this matter is consisting of partonic degrees of freedom (at least in the sense of quasi particles). We know from lattice-QCD calculations at finite temperature and vanishing baryo-chemical potential (i.e., a system with as many bayrons as anti-baryons), which is the situation at very high collision energies reached at RHIC and LHC, that at temperatures of about ##T=150-160 \; \text{MeV}## there is a cross-over transition from a hot and dense hadron gas to such a matter consisting of partons, which is called the Quark Gluon Plasma (QGP).
The difficulty, however, is that we cannot directly observe this (QGP) or even have some box with hot and dense QCD matter at a given temperature and baryo-chemical potential at hand, but the just produced QGP rapidliy expands and cools down, and all we can measure are hadrons (and also leptons and photons) with our detectors, and this gives a snapshot at two stages of this rapid fireball evolution: When the fireball cools down and becomes dilute enough at a certain point the inelastic interactions cease, and the "chemistry" of the fireball is fixed (modulo decays of resonances, which survive longer). Although this is a highly dynamical medium one can describe this chemistry with an astonishing accuracy with a thermal model, assuming that directly at chemical freeze-out the medium is described by an equilibrated medium consisting of the known hadrons, i.e., a hadron-resonance-gas model, and this works over a wide range of collision energies (even down to the low energies at GSI in Darmstadt). At the high energies the chemical freeze-out temperatures are close to the pseudocritical temperature of the cross-over transition between a QGP and a hot and dense hadron-resonance gas. Then the now hadronized fireball cools down and dilutes out further, but elastic collisions are still going on until also these collisions become inefficient (i.e., the mean free path of the hadrons becomes larger than the extension of the fireball), and thus the particles move more or less freely to the detector. Through measuring the energy and (transverse-)momentum spectra of these hadrons one can infer on the state at this thermal freezeout, described by a temperature, various chemical potentials, and a flow field of the medium. At RHIC and LHC you get typical thermal-freezeout temperatures of around 100 MeV.
We infer all this from the fact that the evolution of the bulk of the fireball can be pretty well described by relativistic hydrodynamics, assuming an astonishing short formation time of the plasma (i.e., when the matter reaches local equilibrium and hydrodynamics becomes applicable) of ##\lesssim 1 \; \text{fm}/c##, and the shear-visocisity-over-entropy-density ratio is close to the bound ##\eta/s \geq 1/(4 \pi)##. In fact the NY times ones titled "The most sloshy liquid" about the findings on the QGP at RHIC.