Can you use MLC SSDs for tiered storage or will the number of transactions wear it out?
Yes, MLC SSDs, also known as multi-level cell SSDs, can be used for tiered storage. But whether or not it wears out within the lifecycle of the storage system is dependent on the number of writes, erasures and bit errors, and the amount of wear leveling, error detection and correction, and the NAND die size.
All flash SSDs support a finite number of writes and erasures. This is known as program/erase cycles, or P/E cycles. MLC-based flash has a P/E cycle that typically ranges from 3,000 to 10,000 writes per cell. The way most vendors express SSD lifespan is by total bytes written, or TBW. TBW is calculated by multiplying the P/E cycle by the capacity of the flash SSD. Using this formula on a 200 GB MLC drive with a 3,000 P/E cycle will produce a TBW of 600 TB before the SSD wears out. In other words, that 200 GB drive would require approximately 328 GBs written to it every day for five consecutive years to totally wear it out.
Regrettably, this formula is only a loose approximation and not a definitive one. There are several other factors that reduce the P/E cycle of MLC flash SSDs such as shrinking NAND lithography that increases bit errors (bit errors reduce MLC's useful life), as well as the SSD controller's effectiveness at wear leveling, error detection and error correction.
The good news is that MLC SSD endurance can be improved by reducing the amount of degradation that occurs during a P/E cycle or by increasing the capabilities of the SSD controller to effectively read degraded cells with error correction code. Several MLC SSD vendors have been investing heavily in including investments in digital signal processing, wear reduction management and read-level sensitivity improvements. These high-quality flash SSDs are highly unlikely to wear out when used in tiered storage.
Related Q&A from Marc Staimer
Latency in object stores that stems from a large amount of metadata means the technology is better suited to non-transactional data.continue reading
Eventual consistency in object stores can be an issue because object storage is spread over many nodes and up-to-date data may not always be ...continue reading
HDD failure can put bytes of data at risk. Is multi-copy mirroring or erasure coding the more efficient data protection approach?continue reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.