3.84 TB 2.5-inch SATA mixed-use SSD, ordering code MZ7L33T8HBNA-00B7C
| Capacity | 3.84 TB |
|---|---|
| Interface | SATA |
| Form Factor | 2.5-inch 7mm |
| PCIe Version | Not applicable |
| NAND Type | TLC |
| Endurance (DWPD) | 3 |
| TBW | 21,024 TB |
| Sequential Read | 560 MB/s |
| Sequential Write | 530 MB/s |
| Random Read IOPS | 97K IOPS |
| Random Write IOPS | 60K IOPS |
| Power | 2.3 W |
| DRAM Cache | Yes |
| Power Loss Protection | Yes |
| Encryption | Opal 2.0 |
| Condition | New · Used |
Samsung PM897 3.84 TB, 2.5-inch 7 mm SATA, catalogue code MZ7L33T8HBNA, ordering code MZ7L33T8HBNA-00B7C. 560 MB/s sequential read, 530 MB/s sequential write, 97K random read and 60K random write IOPS at 4 KB, 21,024 TBW — 3.0 drive-writes-per-day over five years — on Samsung V6 128-layer 3-bit TLC V-NAND, with TCG Opal 2.0 and capacitor-backed power-loss protection. Stocked new or as tested pulls with a 1-year warranty from us.
This is the last rung of the PM897 range, and the last rung of the mixed-use SATA ladder Samsung built on this platform. There is no 7.68 TB PM897: the family stops at 3.84 TB, and a deployment that needs more than that per device has to change class rather than capacity. What it does have is 21,024 TBW, which is 11.5 TB of writes a day for five years, and the one cross-generation comparison in the catalogue where endurance is not a variable at all.
The PM897 comes in four capacities — 480 GB, 960 GB, 1.92 TB and this one — and the part-number table lists HBNA as the build code for the top two. Above 3.84 TB Samsung's SATA line continues, but the mixed-use option does not: the 7.68 TB drives in the catalogue are read-intensive parts rated 1.0 DWPD or less, a third of this drive's endurance or below. Anyone who needs 7.68 TB per device and 3.0 DWPD has to leave SATA for SAS or NVMe, and that is a platform decision rather than a drive decision.
Within the family the step from 1.92 TB to this one is pure scale, exactly as every other step in the range is. The performance figures do not move — 560/530 MB/s, 97K/60K IOPS, 105/30 µs, 2.3/3.0 W are identical at every capacity — and the write budget doubles to 21,024 TB alongside the bytes. What changes is the arithmetic around the device: how many of them fit in a chassis, how long a rebuild takes, and whether one of them is an acceptable failure domain.
The density argument is the reason this capacity exists. Twelve of these in a 2U chassis is 46 TB of mixed-use storage drawing 16.8 W at idle; the same 46 TB built from 1.92 TB members takes twenty-four devices, twenty-four idle draws totalling 33.6 W, twice the ports, twice the cables and twice the firmware records to keep in step. Halving the device count is worth more in operational terms than the difference in price per terabyte.
| Product family | Samsung PM897 (V-NAND SSD, data centre / mixed-use) |
|---|---|
| PN | MZ7L33T8HBNA (catalogue code) |
| Ordering code | MZ7L33T8HBNA-00B7C |
| Capacity | 3.84 TB (512-byte sectors, decimal capacity) |
| Form factor | 2.5-inch, 7.00 mm (100.20 × 69.85 × 7.00 mm, max 70 g) |
| Interface | SATA 6 Gb/s, Serial ATA 3.3, ATA/ATAPI ACS-4 |
| NAND | Samsung V6 V-NAND, 128-layer, 3-bit (TLC) |
| Sequential read / write | 560 MB/s / 530 MB/s (128 KB, QD32) |
| Random 4 KB (QD32) | 97K IOPS read / 60K IOPS write |
| Random 8 KB (QD32) | 57K IOPS read / 40K IOPS write |
| IOPS consistency | 95% read / 90% write at 4 KB, sustained |
| Latency and QoS | 105 µs read / 30 µs write at QD1; 99.99th percentile 0.2 ms read, 0.05 ms write at QD1; 0.6 ms and 0.7 ms at QD32 |
| Endurance | 21,024 TBW — 3.0 DWPD over five years (TBW = DWPD × 365 × 5 × capacity) |
| Power | Active read 2.3 W; active write 3.0 W; idle 1.4 W (DIPM off); supply 5 V ±5% |
| Power-loss protection | Yes — capacitor hold-up transfers cached DRAM data to flash on failure; no hold-up time published |
| Data security | AES-XTS 256-bit encryption engine, TCG Opal 2.0 compliant |
| Features | TRIM, NCQ to depth 32, S.M.A.R.T., end-to-end data integrity, RoHS |
| Reliability | MTBF 2.0 million hours; UBER 1 sector per 1017 bits read |
| Environmental | Operating 0 °C to 70 °C (case); non-operating −40 °C to 85 °C; humidity 5–95% non-condensing |
| Code | Type | Condition | Quote |
|---|---|---|---|
| MZ7L33T8HBNA | Catalogue code | New / used | Enquire |
| MZ7L33T8HBNA-00B7C | Ordering code | New / used | Enquire |
The build code on this part is HBNA, the same one carried by the 1.92 TB member; the 480 GB and 960 GB members carry HBLT. It is the field that separates two drives which are otherwise indistinguishable in the hand — same 7 mm case, same connector, same firmware line, same -00B7C ordering suffix. When a mixed pallet arrives, read the label rather than the shape.
The -00B7C suffix is shared across the whole PM897 range and is not interchangeable with the -00A07 or -00B07 codes, which belong to the read-intensive PM893a and to the PM897a build of this range. Those carry different endurance ratings or a different firmware identity, and a PO written against one will not be filled with the other. Quote the full string and we will ship against exactly that.
Put this drive next to the SM883 at 3.84 TB and the usual tie-breaker disappears. Both are rated 3.0 drive-writes-per-day over five years. Both therefore carry exactly 21,024 TBW. There is no endurance difference to trade against price, no write-budget percentage to weigh, nothing to model. The decision reduces to two numbers: 560/530 MB/s against 540/520, and 60K random write against 29K.
That is what makes this the cleanest statement of what the generation change bought. The older drive reaches the same endurance on 2-bit MLC; this one reaches it on 128-layer TLC, writes more than twice as many small blocks per second, and draws 2.3 W on read instead of 2.8 W. If MLC had a remaining advantage at this capacity, this is the pairing where it would show, and it does not: the only lines where the MLC part leads are read latency, 85 µs against 105, and IOPS consistency, 99/97 against 95/90.
Those two are real and they are why the read-serving case still exists for the older part. On a tier that reads far more than it writes, 20 µs of read latency and four points of read consistency are worth having, and the 29K random write never gets exercised. On anything that queues writes, the comparison is not close.
21,024 TB over five years is 11.5 TB of writes every day. Written flat out at 530 MB/s the rating would last about 459 days, which is long enough that the ceiling stops being a planning input and becomes a warranty term. For scale, the whole write load of a busy database node — data files, WAL, temporary tables, backups staged locally — is measured in hundreds of gigabytes a day, not in eleven terabytes.
The figure that actually constrains a 3.84 TB deployment is the rebuild. A full rewrite at 530 MB/s takes about two hours, and for those two hours a mirrored pair is unprotected or a parity set is running without redundancy. That is the number to put in front of anyone proposing this capacity as a mirror: two hours of exposure per device, with every surviving member carrying double the read load throughout. At this size the usual answer is a wider parity group or a distributed scheme rather than a pair, and that is a property of the capacity rather than of the drive.
It is also why the next step up does not exist in this family. A 7.68 TB member would double the write budget to 42,048 TB and double the rebuild to about four hours, and at that point the rebuild dominates everything else about the design. Samsung's answer was to stop the mixed-use line here and offer the larger capacities only as read-intensive parts, where the endurance budget matches what a media tier actually writes.
| Model | Capacity | Sequential read / write | Random 4K read / write | Endurance | Class |
|---|---|---|---|---|---|
| PM897 3.84 TB | 3.84 TB | 560 / 530 MB/s | 97K / 60K IOPS | 21,024 TB (3.0 DWPD) | Mixed-use |
| SM883 3.84 TB | 3.84 TB | 540 / 520 MB/s | 97K / 29K IOPS | 21,024 TB (3.0 DWPD) | Mixed-use, MLC |
| PM893a 3.84 TB | 3.84 TB | 560 / 530 MB/s | 97K / 31K IOPS | 7,008 TB (1.0 DWPD) | Read-intensive |
| PM893 3.84 TB | 3.84 TB | 550 / 520 MB/s | 98K / 30K IOPS | 7,008 TB (1.0 DWPD) | Read-intensive |
| PM883 3.84 TB | 3.84 TB | 550 / 520 MB/s | 98K / 28K IOPS | 5,466 TB (0.78 DWPD) | Read-intensive |
| PM863a 3.84 TB | 3.84 TB | 520 / 480 MB/s | 97K / 24K IOPS | 5,466 TB (0.78 DWPD) | Read-intensive |
| PM897 1.92 TB | 1.92 TB | 560 / 530 MB/s | 97K / 60K IOPS | 10,512 TB (3.0 DWPD) | Mixed-use, half capacity |
The table splits into two blocks and the split is endurance. Two rows carry 21,024 TB; four carry between 5,466 and 7,008, which is a quarter to a third of it. Within the mixed-use pair the sequential figures differ by 20 MB/s and the random write figure differs by a factor of two, so the choice is about the queue rather than the rating. The last row is the same drive at half the bytes, included because it is the only other capacity in the family with the same performance and half the write budget — the trade between it and this one is exactly one variable.
Against the read-intensive group the argument does not need writing throughput at all. The PM893a matches this drive on both sequential figures and on random read, gives up half its random write, and carries 7,008 TBW — a third of the budget. For a tier that fills the drive once and reads it for five years, that third is enough and the price per device is lower. For a tier that writes to it every day, the third is not enough and the drive gets replaced early, which is the most expensive way to buy storage there is.
Every drive carries a 1-year warranty from us against defects in workmanship and function. Verified faulty units are replaced or refunded and we cover return shipping. Stock is held in both new and used condition; the condition you order is confirmed before dispatch, and used units are pull-tested rather than untested lots. Stock is held in our own warehouse and normally dispatches within three to seven working days, with tracking issued at dispatch.
This capacity is quoted rather than listed, because price moves with condition, quantity and batch, and at 3.84 TB the difference between a low-hour pull and a high-hour one is worth asking about explicitly. Send the quantity, the condition you need and the platform the drives are going into, and you will get a quote with lead time the same working day. Matched sets for parity groups are selected from a single batch where stock allows, including firmware revision and wear level.
Tell us who you are — we reply within one working day with a firm quote, stock status and lead time. Volume discounts available.
15.36 TB E1.L 1U long ruler NVMe SSD, PCIe 4.0, ordering code MZEL215THBLA-00AMV
7.68 TB E1.S 1T NVMe SSD, PCIe 4.0, ordering codes MZCL27T6HBLA-00AMV / MZTL27T6HBLA-00A07
3.84 TB E1.S 1T NVMe SSD, PCIe 4.0, ordering codes MZCL23T8HCLS-00AMV / MZTL23T8HCLS-00A07
1.92 TB E1.S 1T (9.5 mm) NVMe SSD, PCIe 4.0, ordering code MZCL21T9HCJR-00AMV