The sharp one-sided tail bound

$X$ has mean 0 and variance 1, and nothing else is known. How large can $P(X \ge 2)$ be? Give the sharp bound and a distribution that attains it, and compare with Chebyshev and with the normal.