In traditional application development, the disk was the main persistence for data storage. The challenge in this method was that, for business logic and application computation, data was transferred from storage disk to main memory, causing huge I/O overhead. Again, after the computations based on the business logic, the data from aggregation, computational, or analytic results was transferred from CPU and main memory to store, or the data was persisted back to storage disk, causing I/O overhead multiple times.
As the following simple illustration shows, disk speed is growing slower compared to other hardware components, while the need for higher performance and speed is increasing day by day: