Protein similarity is the degree that one peptide sequence is similar to another protein sequence. The most commonly used bioinformatic tools to compare peptide/protein sequences are BLAST and FASTA.
Simply speaking, computational biology is the application of computer to the field of biology. In probably early twenty century, the research in biology and related fields had generated/accumulated enormous amount of data, which cannot be efficiently analyzed/processed without the use of computer.
Evolutionary process simply means the process of all organisms evolve from their origin till now. It is the evolution of Darwin's theory. The information preserved throughout this process means DNA, which encodes for protein sequence. Each protein usually contains conserved domains, which is important for the protein functions and that's why it's conserved throughout evolution. Based on that, scientist can identify proteins of the same functions in other organisms, and by modifying conserved domains, they can study the protein functions. Bioinformatics provides a preliminary tool to studying a target gene/protein as nowadays we can do a lot of analyses by computer on one gene sequence such as sequence similarity analysis and protein 3D structure modeling. Before starting wet experiments, computer can save us plenty of time in research, and the generated preliminary data usually point out a clear and correct research direction.