With space to change time, the memory database is more than fast

In recent years, with the fall in memory prices, memory database is becoming more and more popular, the performance of memory database also has a leap-forward development. However, many developers only think of the speed when considering the memory database, which undoubtedly ignores the real potential of memory technology.

Memory provides more than just a faster speed, it is also a way to store space for time. Many developers have some doubts about the use of in-memory databases, and here are one by one notes:

1. The in-memory database runs relatively fast. Is there anything else besides fast? We know that the memory reads and writes much faster than the hard disk, so the table is loaded into memory to achieve high read and write speed. The size of the virtual memory is limited by the swap partition, and typically the size of the swap partition is one to twice times the amount of memory, so the size of the data that can be loaded in memory is also constrained. The amount of time it takes to process data when the volume of data being loaded is close to the swap partition capacity. Even now that the memory is cheap, it is still not enough relative to the amount of data to be processed.

**2.32 bit to 64 bit just recompile the code? The **64-bit architecture allows the system to break through the limitations of 4GB memory. If there is no 64-bit architecture then the memory database will be much less applicable. Since the 64-bit CPU was released, many developers are talking about software that was developed under 32-bit architectures to recompile to 64-bit. 2^64 means an almost limitless amount of addressing space, so there is also an infinite possibility of space-time-swapping algorithms.

3. What are the limitations in virtual? The term virtual memory was first introduced in 1970 by Peter Denning, which divides the memory used by the application into parts to enable efficient use of computer memory, which also has the potential to exchange portions of the physical memory to the hard disk. However, due to the limited capacity of the swap partition, we still need to consider the impact of memory on the application.

Only 48 bits are used in 64-bit addressing space, but it is sufficient for big data applications. When memory is no longer a scarce resource, we need new ways to solve the problems encountered in big data projects.

With space to change time, the memory database is more than fast