It has been hypothesized that the main advantage of sexual reproduction -- which causes every member of a species to be genetically different -- is resistance to microbes and other parasites.
If you examine asexual reproduction: you get a "tree of evolution". Each node in that tree is slightly "more evolved" than its parent on average (natural selection prunes the devolved ones more than the evolved ones).
However, each branch in the tree has to evolve every single feature independently from its sibling branches. You only inherit evolutionary enhancements of your own ancestors.
Assuming a tree of depth N: each node in the tree has only O(N) ancestors, and thus, about O(N) evolution experienced in each node.
With sexual reproduction, you also recombine the evolutionary advancements of two ancestors into new nodes. It is no longer a tree, but a DAG of nodes (Directed Acyclic Graph). Each node now has O(2^depth) ancestors. Evolution can recombine features that took a long time to develop into a new baseline, and evolve from there.
This is an exponential increase in evolution speed -- and I believe that is so incredibly huge of an advancement, that I predict that even species we believe are incapable of sexual reproduction, are probably sexually reproducing (e.g: exchanging DNA in some way) because it is too huge of an advantage to give up.
Lastly, I find that it is interesting that there's a similarity between sexual-based evolution process to one of the frameworks we use for parallel computing: "Map/Reduce". Take a single node in the tree, and just "Map" (mutate & select) over it multiple times, you get slightly better nodes. Now combining the mutated & selected nodes together into descendants is much like "Reduce". "Map" is not a useful parallel computation, but "Map & Reduce" is.
The disadvantage to sex is you need vary similar organisms so your N^2 ancestors becomes (species population size) * number of generations. Where direct DNA exchange can cross DNA from far less compatible organisms (~total life population size)
See wikipedia: http://en.wikipedia.org/wiki/CCR5_delta_32#CCR5-.CE.9432
There's another gene with a variant that confers similar resistance to noroviruses, also called stomach flu.
Why shouldn't there be people who are immune to HIV?
HIV immunity is not at all analogous to that.
You see, there are receptors on white blood cells - which are involved in lots of things the immune system does, if you're interested you can find an explanation elsewhere. Normally, HIV uses the CCR5 receptors to enter white blood cells, in which it replicates. However, in some people the genes that generate that receptor are mutated, with a small portion being deleted. It still works for most of the things the immune system does, but HIV is no longer able to enter cells through it (I'm not entirely sure if it's unable to enter or just unable to enter anywhere near as well; the former would suggest total immunity while the later would be partial immunity).
See also http://en.wikipedia.org/wiki/CCR5