Search code examples
mathstatisticsusage-statisticstrendtrendline

How to calculate the lifetime of the use of a database?

Can you please help me with this problem?

You have the following information:

Name     Date     Size_Total  Size_Free  Size_Used
 X    20/05/2019     50MB       40MB       10MB
 X    21/05/2019     50MB       35MB       15MB
 X    22/05/2019     50MB       26MB       24MB
 X    23/05/2019     50MB       24MB       26MB
 X    24/05/2019     50MB       22MB       28MB
 X    25/05/2019     50MB       17MB       33MB
 X    26/05/2019     50MB       15MB       35MB

These data are extracted from a database daily for monitoring. What statistical function can I use to determine how many days are left to the database to run out of space according to its use?

I appreciate the help.

Thank you so much

Solution

  • A simple solution would be to use linear regression to come up with a linear model of the database size.

    Data Input

    Using this input generates the following output.

    Linear Regression Result

    The equation of the slope of the line is what we're looking for: y = 4.107x + 8. We can use this to find when the database will reach its maximum size by determining where this function intersects with y = 50, the size of the database. This second equation is a horizontal line because the maximum size of the database does not vary with respect to x, time, which is the whole problem.

    Looking for the intersection yields the following result:

    Graph of the Answer The Result

    Remember, however, that the answer, 10.2264, includes the days that have already gone by. We had seven days worth of input, so since our model predicts the database will fill up on day ten, we have roughly two days and some change. This is not a hard figure though, because remember that the database size is increasing irregularly and we could much more or much less time if our base data was not representative of the true database use, or if external factors that directly impacted database use changed.

    Note that I did not include computations in the answer because I'm focusing on the high-level concept. I used the first linear regression calculator a quick google search pulled up and then plugged in the equation into Wolfram Alpha, so please actually do this yourself to verify the numbers are correct.