Most of the common DBMS (ORACLE, Microsoft SQL Server, MySQL, DB2, etc.) use disk storage as run time memory. When data becomes necessary from database, DBMS reads them from disk and loads it to RAM (Random Access Memory). This approach uses small amount of RAM, but DBMS efficiency is relative poor. In this work was created prototype of DBMS. This prototype is read-only DBMS and holds all database data in RAM. The purpose of this work is to compare efficiency of common DBMS against created prototype. For this prototype was developed and optimized bunch of data searching, sorting, grouping, joining algorithms. All of these algorithms are based on the main prototype data model idea: database table stores all data into table tree. Prototype and other DBMS were tested with TPCH test, which consists of very different 22 SQL queries to test DBMS efficiency. The test result produced good results for prototype: prototype was 30-40 times faster against ORACLE on complex joining queries and 100-200 times faster against ORACLE on simple joining and searching queries.