don't need enterprise levels of endurance in a 2013 era rig that gets used a few times a month.
Enterprise SSD "small deals"
- Thread starter ca3y6
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
Fair enough, but when you buy used, you are not the only one who wrote to the drive.
I received my 12 drives today and all of them are 2020 April/May/June with the exception of 1 2019 November, and all have between 35,000-43,000 POH, all have POC under 100 and writes are below rounded up to closest GB or TB
4 Drives
GBW - 2, 2, 4, 131
8 Drives
TBW - 2, 2, 4, 29, 39, 164, 164, 1186(not a typo)
Wow that one drive put in the work. Should still have a few petabytes left though!
Yes it did.
For those of you who bought some of these I just finished testing all these drives and 10 of them fail via Scrunity for a high amount a Command Timeout errors and 2 of them get a clean bill of health. From my research there appears to be a bug in the firmware MP32 and earlier and was resolved in firmware MP33 so if you purchased some I suggest updating the firmware to MP33. Micron//Crucial Storage Executive says MP32 is the latest but it appears to not be up to date.
MP33 Firmware
Last edited:
Not super related to this thread, but anyone mind sharing how they test drives after they get them? I got 2 Micron 5300 Max's a year or so back and they had 0.2 and 3.2 PBW, yes a 3PB difference from the same buyer (rated for ~24PBW). Both were literally the same POH ~19k so I have no idea what this madman was doing with them, but they both maintain ~500MB/s. I didn't think much about slowdowns because my experience with SATA SSDs has only been with new drives and in my regular use I've never noticed a significant (like >50-100MB/s) slowdown before death of consumer SATA SSDs. A drop of 10-20% even I'm fine with, but I also have one significantly worse.The Micron 7300 I got arrived with 2PB on them that's less than 10% of what they are rated for.
I only started thinking of this now because I found one I bought is extraordinarily slow (still can return but I have data on it and I'm too lazy and the data isnt moving
). It's a Samsung 863 running at 100MB/s and it had only used 5% of it's write endurance and had 34k POH. Does only PBW factor into write slowdowns or do you consider POH as well? I've never really done more than checking SMART is good. Any suggestions?First is Chinese "Quality" ... at best cheap at worst Fraud (faked Capacity) making it look like a Samsung while it clearly is NOT.Sorry, I might have taken a bad example
Its a 4tb m2 for $39
Or this 2TB SSD 2.5" for €89 (which might not be such a good deal btw):
2x Osram H7 Classic 64210 CLC Lampe 12V 55W 64210CLC Autolampe Glühlampe Birne | eBay
2x Osram H7 Classic 64210 CLC Lampe 12V 55W 64210CLC Autolampe Glühlampe Birnewww.ebay.com
Second is 2TB for 90 EUR from a very low-end Brand (Intenso), so hardly a Deal at all.
You are 100% correct, I just finished updating all of ones I got to MP33
I typically test mine with the manufacturers tools designed for the task and then test them using Scrutiny which provides a nice UI that shows more info than a standard SMART test.Not super related to this thread, but anyone mind sharing how they test drives after they get them? I got 2 Micron 5300 Max's a year or so back and they had 0.2 and 3.2 PBW, yes a 3PB difference from the same buyer (rated for ~24PBW). Both were literally the same POH ~19k so I have no idea what this madman was doing with them, but they both maintain ~500MB/s. I didn't think much about slowdowns because my experience with SATA SSDs has only been with new drives and in my regular use I've never noticed a significant (like >50-100MB/s) slowdown before death of consumer SATA SSDs. A drop of 10-20% even I'm fine with, but I also have one significantly worse.
I only started thinking of this now because I found one I bought is extraordinarily slow (still can return but I have data on it and I'm too lazy and the data isnt moving
If slow, might be worth doing a TRIM, updating the firmware.Not super related to this thread, but anyone mind sharing how they test drives after they get them? I got 2 Micron 5300 Max's a year or so back and they had 0.2 and 3.2 PBW, yes a 3PB difference from the same buyer (rated for ~24PBW). Both were literally the same POH ~19k so I have no idea what this madman was doing with them, but they both maintain ~500MB/s. I didn't think much about slowdowns because my experience with SATA SSDs has only been with new drives and in my regular use I've never noticed a significant (like >50-100MB/s) slowdown before death of consumer SATA SSDs. A drop of 10-20% even I'm fine with, but I also have one significantly worse.
I only started thinking of this now because I found one I bought is extraordinarily slow (still can return but I have data on it and I'm too lazy and the data isnt moving
Personally I collect and monitor all my smart data centrally (I want to be notified if for instance uncorrectable errors increases, or if the health unexpectedly reduces, happened to me when I was underestimating the writes on some drives ). I also run crystaldiskmark. If it's a retail drive and I am worried about counterfeits , I attempt a full disk write.
Not sure what those generic drives are worth but WL OEM 7.68TB - 1 DWPD SAS 12Gb/s 2.5" Enterprise SSD $200
This might be a pricing error because they also have listings for the same drive (serverpartdeals) for $350 unless they're tanking the price White Label OEM SAS 12Gb/s 2.5" Enterprise SSD - 7.68TB, 15.36TB | eBay
They also have 90 day warranty from serverpartdeals and a different version with a 1 year warranty for $380. I would buy if I needed them.....but maybe I still want 2 or 4... or shit am I going to be buying 8 drives and an HBA this weekend?
They also have 90 day warranty from serverpartdeals and a different version with a 1 year warranty for $380. I would buy if I needed them.....but maybe I still want 2 or 4... or shit am I going to be buying 8 drives and an HBA this weekend?
Not a Deal (at least for me) but a Remark about the Deal that I managed to grab thanks to @ca3y6 a few Months back about 3 x Intel (Oracle) P4608 6.4 TB / Piece for 399 EUR plus Shipping (~ 142 EUR / Piece Shipped).
There is now another Auction for it from the same Seller, but the same Bundle is set to a "Buy Now" Price of 699 EUR which is almost double what I paid !
So yeah, it might be that the Sellers have Pallets of it and they bought them for like 50 EUR / Piece, but maybe they put the wrong "Buy it Now" Price back then at 399 EUR.
There is now another Auction for it from the same Seller, but the same Bundle is set to a "Buy Now" Price of 699 EUR which is almost double what I paid !
So yeah, it might be that the Sellers have Pallets of it and they bought them for like 50 EUR / Piece, but maybe they put the wrong "Buy it Now" Price back then at 399 EUR.
there is only one at this price so your concerns about buying 8 might be premature
1.5TB Optane D4800X on Yahoo Japan auction for $140: Item Details
seller reports only 900gb detected but it's probably just partitioned
seller reports only 900gb detected but it's probably just partitioned
Besides that it's like 50% more expensive / TB than some of the other Deals, WHERE and HOW would you even install it ?
Judging from the Picture it's like 4x longer than a Normal 2.5'' SSD Drive
.Smartctl for extracting the data to a json file, then my own code for the rest. Also needs to work across OS (ubuntu, windows, synology).
I store smarts daily in a database. I separately log to file lots of other metrics: drives usage, temperature, same for CPU, ram usage, network card usage, etc, every 5 minutes. Useful to see how long a problem occured, or if a disk is filling up, or a server is out of RAM, etc.
Also for drive pools (zfs, storage space, md), I keep track of aggregated writes, disk topography (which disk connected to which port of which HBA), etc
Also for drive pools (zfs, storage space, md), I keep track of aggregated writes, disk topography (which disk connected to which port of which HBA), etc
I am actually writing a tool in Rust to export metrics from Linux systems to Prometheus. This will also include latency metrics from the disks. It is still work in progress, and will be open-sourced. The metrics are stored in 1 second interval, so you get high data resolution.
I think that atleast 100 times a day. I envy people who’s minds are content just “being”.Way too many Projects compared to the Time available
Slow deal week. Next time $100 4tb SAS drives pop I’ll buy without trying to haggle.
LOT OF 2 - Toshiba PX05SRB384 3.8TB SAS 3 2.5" SSDs $275 ($137.5 each)
Didn't try and I personally didn't bother. I think I reported the FW Version a few Pages back, let me know if it's not there then I'll probably try to dig it up.
Right now it's a bit left and right everything here so for now the Server is not running, I just did a quick test on it.
I think there was a Oracle Firmware in one of the Threads here, but also not sure if it was "recent" or if the one I got was basically the same anyways.
EDIT 1: probably NOT possible to flash Intel Stock Firmware, Oracle is known to lock down their Devices pretty tightly, so it's basically like this until it dies (hopefully in a VERY LONG Time
).
There is no point reformatting them to 4k. IO sizes are dynamically aligned. If you write small files you will save space. If you write large files, you will have same performance. If you use PostgreSQL you never write smaller blocks than 8kb, if you use MySQL you never write smaller blocks than 16kb.