[throwaway2037]

I checked the specs here: https://www.micron.com/content/dam/micron/global/public/prod...

The interface looks equiv to 4x PCIe 5.0.

    > Sequential read (MB/s): 13,700
    > Sequential write (MB/s): 2,700

That is pretty awful write performance. Does anyone know more about this? I assume all of these hyperdense SSDs suffer from the same drawback. Also, I heard that the E3.L interface can support up to 16x lanes, but there are no practical commerical products at this point.

[Aurornis]

Consumer and data center drives play by different rules. The super high write speeds you see for consumer SSDs are usually achieved through tricks like using sections of the drive as a high-speed buffer and then using a background process to rewrite the data into the drive's high-density NAND storage during downtime. They can also use caching techniques that aren't resilient to power loss. They might allow burst performance that heats the drive up until it throttles.

This is all fine and even desirable for a consumer who will only be writing for at most a minute or two, but with a 245TB server drive you need to assume the performance will be needed constantly. They target sustainable and predictable performance.

[justsomehnguy]

Extremely dense QLC chips. Still it's 2700-3000MByte, ie ~3GByte/second.

What should worry way more is DWPD which is abysmal... on the first glance. But if you punch it in the calc it still would take ages to wear it out.

                    SSD #1    SSD #2     SSD #3
    Capacity (GB)   245000    245000     245000
    Warranty (yr.)  3         3          3
    DWPD            0.3       1          0.075
    TBW (TB)        80482     268275     20121
    TBW (PB)        80.483    268.275    20.121
    PBW             80.483    268.275    20.121
    GB/day          73500     245000     18375
                
    Time period Average host-side write data rate (MB/s) needed for reaching DWPD value within specified time period
    8 hr.           2552.08   8506.94    638.02
    12 hr.          1701.39   5671.30    425.35
    24 hr.           850.69   2835.65    212.67

https://wintelguy.com/dwpd-tbw-gbday-calc.pl

[_zoltan_]

DWPD was the boogey man 10 years ago. everybody worried about it.

now, nobody cares. I have over 500 NVMe drives in our deployment and the drive deaths are not due to wear.

[throwaway2037]

    > I have over 500 NVMe drives in our deployment and the drive deaths are not due to wear.

This is high quality anecdata. What is the root cause of most of your drive deaths?

[markhahn]

then either your drives are overprovisioned or read-mostly.

it's not that hard to hit 300 cycles on flash.

[_zoltan_]

it's all 4TB or larger plus the drives do wear leveling internally.

[sheepscreek]

A more convenient (and dare I say, faster) tape drive replacement for backups? They do make a good point, it would take 10*24TB drives working in the worst raid configuration to even come close to these speeds.

[voxelghost]

65 hours to restore a full backup

[xattt]

Yes, but with all that data, how much heavier does it get?

[jurgemaister]

2.231705*10^-13 gram

[hebelehubele]

:)

A single speck of dust could throw off that measurement (~ 1.6 x 10^-7 grams)

[didgetmaster]

It depends on the data. I heard that ones are heavier than zeros.

[thrill]

Yes, this why drives come in the mail pre-wiped - it saves on shipping.