<![CDATA[Abstract The development of social network applications has growth so fast with various applications supported by smart tools as multiplatform to used it. Several applications that used by user are the chat and media social app. However, there is problem is the number of users accessing the chat service is large. This condition makes communication into database server can be stop and loss of data. It is caused by excessive load received by a single server database. Therefore, this study designs a database server to solve a lot of users of communication server and storage capacity into database server more than one. This is important because to increase availibiliy of services for each users request. This study used a method of distribution of communications service request for each database server. Distribution of communication service request used a load balancing method and HAPRoxy with scheduling method. There are two algorithms in scheduling method are round robin and least connection algorithm. Both of algorithms were evaluated. compared, and used in database server more than one. The result shown average of least connection algorithm has value of response time 32.421 ms is smaller than round robin algorithm 35.813 ms. While on the throughput, least algorithm has big value 211.267 Kb/s than round robin algorithm has value 210.298 Kb/s. The result shown the number of packet least connection algorithm are big and better than round robin algortihm with load balancing implementation for distribution of server communication on database server more than one able to solve a lot of services communication. Keywords: load balancing, database server chat, least connection, round robin, response time, throughput]]>
