Over time, systems connected to databases slow down. This is usually due to the increase in the amount of data stored in individual tables, counted even in the billions of records. Nevertheless, there are methods for making the speed of the system independent of the number of records in the database. One of these ways is table partitioning. When used correctly, the solution can ensure efficient operation of very large databases even after several years. However, not everything is predictable because of some undesirable phenomena become apparent only with a very large amount of data. The article presents a study of the execution time of the same queries with increasing number of records in a table. These studies reveal and present the timing and circumstances of the anomaly for a certain number of records.