Skip to Content

While querying the web for information about the CAP theorem  I found an interesting paper titled “The five-minute rule twenty years later, and how flash memory changes the rules” by Goetz Graefe.

 

The “five minute rule” is the time interval between accesses the same data for which is does not matter if the data is stored on disk or in memory (cache). Consequently, if the data is accessed more frequently it should be stored in memory (cache) and if data is accessed less frequently it should be stored onto disks

 

In 1987 the five minute rule was valid for a data sets of 1 KB. In 1997 the five minute rule was found to be valid for page size of 8 KB. In 2007 the five minute rule was valid for a data sets of 64 KB. For these calculations the prices for 1MB of RAM were estimated to 5000$ in 1987, 15$ in 1997 and 0,05$ in 2007.

 

Another interesting aspect mentioned in the paper is that small data sets can be kept in memory for a very long time now (2007) without having to swap them out of the main memory because the break even point is a very long time interval, longer than the typical time between two snapshots.

To report this post you need to login first.

2 Comments

You must be Logged on to comment or reply to a post.

  1. Clemens Li
    Thank you for the idea!

    The most difficult part will be to know which data is accessed more than every five minutes, especially in a multi-user environment.
    With shared objects, ABAP provides a powerful but not easy to implement way of providing in-memory-data.

    Good, in theory, though. If you look at millions of code lines that should have been changed already…

    (0) 
  2. An interesting article, I must admit. I will have a look at my coding from now on. Singleton will be needed. BTW: I am still single.
    (0) 

Leave a Reply