Thank you for your question, which allows me to expand on a couple of points. You refer to the NAND flash’s limited number of program and erase (PE) cycles as a reliability concern, but this is not really the case in practice. Yes, NAND has a limited number of PE cycles, but a managed NAND device like NANDrive handles this to ensure that the NAND is used effectively. Limited PE cycles is not about reliability, because if a block fails early it is replaced from the spare pool and its data is preserved. Consideration of the usage model for NANDrive during the design phase will ensure that a correct device is selected to meet the application requirement for PE cycles so there is no reliability impact. Thank you too for the article reference you posted. The thrust of my article is for embedded systems, and not aimed at consumer or data centre applications. The embedded example I used is more about read intensive applications and code execution from RAM. The case for reliability concerns in HDDs is largely a mechanical one in embedded designs, which require robust mass storage. There is ample evidence from my experience of working with embedded customers and understanding their requirements. These customers report that the mechanical drive is a large point of failure in their systems, and for that reason solid state solutions are becoming more prevalent. Embedded industrial applications tend not to use HDDs because of these mechanical reliability concerns.
" HDDs are bulky and power hungry.. These characteristics, combined with reliability concerns, make them impractical for use in embedded systems despite the fact that HDDs have a very low cost per gigabyte (GB)" - Most peoplr associate reliability concerns with SSDs, and their limited number of Program/Erase cycles. Can you provide a technical source article or JEDEC for that comment ?