There's more...

Additional optimization techniques include reducing unnecessary precision in decimal numbers (which tends to reduce the uniqueness of values) as well as using integers versus text whenever possible. Finally, implement the table in this recipe as shown in the following manner:

R10_Table3 = 
    VAR __Table = 
        ADDCOLUMNS(
            ADDCOLUMNS(
                GENERATESERIES(1,1000000,1),
                "TimeStamp",NOW()+[Value]/24/60/60
            ),
            "Year",YEAR([TimeStamp]),
            "Month",MONTH([TimeStamp]),
            "Day",DAY([TimeStamp]),
            "Hour",HOUR([TimeStamp]),
            "Minute",MINUTE([TimeStamp]),
            "Second",SECOND([TimeStamp])
        )
RETURN
    SELECTCOLUMNS(
        __Table,
        "Year",[Year],
        "Month",[Month],
        "Day",[Day],
        "Hour",[Hour],
        "Minute",[Minute],
        "Second",[Second]
    )

You will see that this version of the table only takes 34 KB of storage within the model.