What Is SSD TBW?

TBW, or terabytes written, is the endurance rating printed on an SSD’s datasheet. It states how much data the manufacturer warrants can be written to the drive before the NAND flash cells are expected to wear out. A 1 TB drive rated at 600 TBW is warranted for 600 TB of host writes, or until the warranty period ends, whichever comes first.

The limit exists because flash memory wears as it is used. Each program/erase (P/E) cycle stresses the thin oxide layer that traps charge in each cell, and after enough cycles the cell can no longer hold data reliably. Hard drives have no equivalent write limit, which is why TBW is specific to flash storage such as solid-state drives and eMMC storage.

Manufacturers rate endurance using the JEDEC standards JESD218, which defines the test method and acceptable error rates, and JESD219, which defines the client and enterprise workloads written during the test. Two drives with the same TBW figure may still have been rated against different workloads, so the datasheet footnotes matter.

TBW comes from three numbers: the drive’s capacity, the P/E cycles its flash is rated for, and the write amplification factor (WAF), which is how many bytes the drive actually writes to flash for each byte the host sends. In simple form, TBW ≈ capacity × P/E cycles ÷ WAF.

DWPD, or drive writes per day, expresses the same endurance as how many times the full capacity can be written every day of the warranty. DWPD = TBW ÷ (capacity × 365 × warranty years). A 1 TB drive rated at 600 TBW with a five-year warranty works out at about 0.33 DWPD. Client and read-intensive drives are typically around 0.3–1 DWPD, mixed-use enterprise drives around 3 DWPD and write-intensive drives 10 DWPD or more.

TBW is the better figure for comparing drives of the same capacity. DWPD is easier for planning, because it scales with capacity and maps directly onto a daily write budget.

The NAND type sets the starting point. Storing more bits per cell raises capacity and lowers cost but cuts the number of P/E cycles each cell survives.

Figures are typical ranges and vary between flash generations and manufacturers. Our comparison of QLC vs TLC covers the trade-off in more detail.

Work from the application’s daily writes, not its capacity. Measure or estimate how much data the system writes per day, multiply by the years of service you need, then add a margin for write amplification and growth.

For example, an edge computer logging sensor data and short video clips might write 50 GB a day. Over seven years that is about 128 TB of host writes; doubling it for write amplification gives roughly 256 TB. That fits within the roughly 300 TBW typical of a 512 GB TLC drive, while a 1 TB drive at around 600 TBW leaves more headroom. A system recording several continuous video streams can write terabytes a day and needs a write-intensive or pSLC drive instead.

Operating system settings matter too. Moving logs and temporary files to RAM, limiting swap and using a write filter, as in Windows IoT Enterprise or a read-only Linux root file system, can cut writes dramatically on a system that mostly reads.

SSDs report their own wear through SMART health data. NVMe drives expose a standard “Percentage Used” value and a lifetime count of data written, kept in blocks of 512,000 bytes. SATA drives report wear and total writes through vendor-specific attributes, often labelled “Total LBAs Written” and a wear-levelling or media-wearout indicator. Tools such as smartctl read these on Linux and Windows, and many industrial drives add their own monitoring utility.

TBW is not the same as reliability. Endurance tells you when the flash is expected to wear out; MTBF describes random failures across a population of drives during their working life. Our explainer on what MTBF means covers the difference.

We choose the flash type, capacity and endurance rating for every SSD we fit as standard, sized to the system’s expected writes and operating temperature, then image and validate the system before it ships.