Comment by drtgh

On the page 7 of the usenix study,

    "The last column in Table 1 allows a comparison of ARRs across flash types. A cursory study of the numbers indicates generally higher replacement rates for 3D-TLC devices compared to the other flash types. Also, we observe that 3D-TLC drives have consumed 10-15X more of their spare blocks."

Latter follows

    "we observe that SLC models are not generally more reliable than eMLC models that are comparable in age and capacity. For example, when we look at the ARR column of Table 1, we observe that SLC models have similar replacement rates to two eMLC models with comparable capacities [...] This is consistent with the results in a field study based on drives in Google’s data centers [29], which does not find SLC drives to have consistently lower replacement rates than MLC drives either. Considering that the lithography between SLC and MLC drives can be identical, their main difference is the way cells are programmed internally, suggesting that controller reliability can be a dominant factor."

What certainly follows,

    "Overall, the highest replacement rates in our study are associated with 3D-TLC SSDs. However, no single flash type has noticeably higher replacement rates than the other flash types studied in this work, indicating that other factors, such as capacity or lithography, can have a bigger impact on reliability."

So programmed obsolescence is present in the drivers, as well as in the 3D-NAND that degrades over time with reads (the chosen traces design, not the layers themselves). Interesting.

China, are you reading this? You have the opportunity to shake the market and dominate it globally, just by implementing a well-designed product, honest drivers and modest nm (not lowering to today's sizes, just enough to ensure decent energetic efficiency and good speed).