What is the compelling need for introducing a theory of $p$-adic integration?
Do the existing theories of $p$-adic integration use some kind of analogues of Lebesgue measures? That is, do we put a Lebesgue measure on $p$-adic spaces, and just integrate real or complex valued functions on $p$-adic spaces, or is something more possible like integrating $p$-adic valued functions on $p$-adic spaces? What is the machinery used?
Then again, does the integration on spaces like $\mathbb C_p$ give something more than the usual integration in real analysis? I mean, the integration of complex valued functions of complex variables, or more precisely holomorphic functions, is much a much more interesting topic than measure theory. Is a similar analogue true in $p$-adic cases?
I have also seen mentioned that Grothendieck's cohomology theories like etale cohomology, crystalline cohomology etc., fit into such $p$-adic integration theories. What could possibly be the connection?
I must say I do not know anything much about it; but the following is the general idea.
Note that in the case of cohomology theories for smooth projective varieties over $\mathbb C$, there are various canonical isomorphisms between de Rham cohomology, Betti cohomology, etale cohomology, etc.. For instance the isomorphism between de Rham cohomology and Betti cohomology involves integrating differential forms over homology classes. Actually the situation is a bit more intricate; the de Rham cohomology decomposes into various other spaces via the Hodge decomposition. This whole stuff is all done over the complex numbers.
In the $p$-adic case, an analogous theory is built, called the $p$-adic Hodge theory. Some details are available at the wikipedia page. There is an isomorphism between algebraic de Rham cohomology and $p$-adic etale cohomology, as per the Hodge-Tate conjecture. There is another conjectural functor from the algebraic de Rham cohomology to $p$-adic etale cohomology, which is called Grothendieck's mysterious functor. A theory for this was constructed by Jean-Marc Fontaine and it was applied in various situations such as the study of $p$-adic Galois representations.
The theories of $p$-adic integration are useful in connection with such investigations. Pierre Colmez has a theory and so does Robert Coleman. Coleman's homepage has a course on this material. Also see Perrin-Riou's Asterisque volume [3]:
[3]: Fonctions L p-adiques des représentations p-adiques, Perrin-Riou, B.