"Barycenter" is really just a fancy term for the center of mass of a system. You can compute it quite easily if you know the positions of all the objects in a system at a given time. This is usually quite handy, at least when dealing with either a pair of similar-mass objects, such as a binary star, or in systems where one object is gravitationally dominant and most other objects orbit it, like in a planetary system.
A barycenter isn't really "fixed", because no reference frame is absolute; we know this from the principle of relativity, which discards the idea of privileged inertial reference frames. However, when discussing orbital mechanics, it's often quite convenient to use barycentric coordinates. In a binary star system, for instance, the barycenter is a focus of the orbits of both bodies, and so it is an obvious choice for designating as a reference point, as we can then calculate the stars' positions.
Now, things get complicated if we look at more complex systems. For instance, Earth doesn't really orbit the Solar System's barycenter; it orbits the Sun-Earth barycenter - which is, for most purposes, the center of the Sun. Finding the barycenter of the Solar System isn't too important if you're computing Earth's orbit approximately (for more exact calculations, you might want to take other bodies into account, namely, the giant planets).
Note that the equation for the center of mass of two bodies of masses $m_1$ and $m_2$ and positions $\mathbf{x}_1$ and $\mathbf{x}_2$ is
$$\mathbf{x}_{\text{cm}}=\frac{m_1\mathbf{x}_1+m_2\mathbf{x}_2}{m_1+m_2}$$
and if $m_1\ll m_2$,
$$\mathbf{x}_{\text{cm}}=\frac{m_1\mathbf{x}_1}{m_1+m_2}+\frac{m_2\mathbf{x}_2}{m_1+m_2}\approx\frac{m_1}{m_2}\mathbf{x}_1+\frac{m_2}{m_2}\mathbf{x}_2\approx\mathbf{x}_2$$