Internet users are gradually increasing, with many needs or information that must be searched through the internet now no longer a problem with the existence of information service providers such as websites. The problem comes when users with very much speed accessing a website at the same time or concurrent access becomes an obstacle. This can be handled using load balancing methods and also server virtualization. Load balancing is a method or technique of distributing traffic loads on two or more connection lines equally. So that traffic runs optimally and stable and overcomes c10k problems, namely optimizing network sockets when receiving requests reaching 10 thousand connections at the same time. Server virtualization is a method used to make several physical servers into one physical server container, to divide the computational load in one physical server container. The results obtained by testing the access speed simultaneously using the webserver stress tool application with different user simulations are 25 users, 55 users, 75 users, 100 users, 125 users, 250 users and 1000 users. The results of the access speed test are again tested for the average difference using the ANOVA test, which allows researchers to test the comparison hypothesis on the average of more than two groups of round robin, weighted round robin and leastconn algorithms with the results, that at the moment 25 and 55 users of round robin algorithm are better than weighted round robin, but when 75, 100, 125, 250 and 1000 users of weighted round robin algorithm are better and optimal than the round robin algorithm when accessing thousands of data from web servers and algorithms leastconn did not provide good performance as evidenced by the seven different tests and simulations, but it did not give an average result that is less than the round robin and weighted roundrobin algorithm. Of the seven simulations that have been carried out. So the use of weighted round robin algorithm is better than round robin and least connection.