They may work in Windows, I didn't mean to say it was only usable on Linux, there's a big thread on here that might have that information.
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.
That's a relative term, the write lifetime of any Fusion IO is still absurdly long, performance is still pretty solid, and I believe you can manually set the usable capacity to 2.6TB to get the same effect as buying the write intensive version.
In enterprise SSD parlance, read intensive is often 1DWPD, mixed use is 3DWPD and write intensive is like north of 10DWPD. Even 1DWPD is a massive amount of write for any normal usage.
They all use the same NAND, you can easily write 32PB to the 3.2 and double that to the 6.4TB model no problem... that's just the warrantied wear life as well, it will go much longer.
You can do 7-10 DWPD on all the cards just fine in most cases. It really comes down to your app any any write amplification. Even then, you're likely fine.
That's just marketing. They were pre-provisioned with less unused space for higher capacity, which can put more pressure on the groomer under heavy write workloads and will reduce performance. That provisioning line is fully user adjustable though and you can set it to whatever you want. There's no difference in the physical product.
Tell me more... I haven't actually seen one die on its own in my years of working for Fusion-io or even maintaining the cards to this date in my home lab.
Last edited:
Yep, before the COVID i had two Fusion io cards, one with 3.2tb and other with less (2tb i believe ?!) and both "died" (actually zombified)
The 3.2tb started reporting as 1tb, then 500 mb and then 1gb.
The 2tb simply disappeared from the pc, simply died.
Both pcie, both with large capacitors.
They both ran hot and IIRC heat probably killed them.
Current Prices on eBay Europe have shot up quite a Bit.
Whereas before you could get a 14TB used HDD for 100 EUR, now you can "win" an Auction for 8TB at around 100 EUR (maybe 90 EUR).
Whereas before you could get a 14TB used HDD for 100 EUR, now you can "win" an Auction for 8TB at around 100 EUR (maybe 90 EUR).
Yep, didnt had much choice back then, here in Brazil 40° is the NORM, so everything ends dying to heat.
Computers are comfortable with much warmer temperatures than humans are. If you have sufficient heat transfer between the card and the ambient air, the card will be quite happy at 40C. But not at 90C. Different components also have different temp specs.
That shouldn't be ... The HDDs shouldn't block the Operating System Boot Process.
The only case where I had an HDD Pool taking FOREVER to Boot it this damn Server where for whatever Reason, one HDD doesn't want to spin up correctly. I changed Backplane, Cables, HBA and HDD. Nothing. It still refuses to boot. Might be some Mechanical Tolerances, I don't know
.In that Case, if the HDD is plugged in, it will easily take 15-30 Minutes for it to try to boot.
Are you using ZFS or MDADM ?
Did you check Systemd Logs ? It could be that you are trying to import the Pool at Boot Time (via e.g. ZFS Cachefile).
If you have something in
/etc/fstab causing a Hard Dependency, adding nofail,x-systemd.automount as Mount Options should help quite a bit.
The NAND certainly doesn't care about the temps, but that FPGA does and needs ~300LFM of airflow.
Probably not the worst deal to miss. I bought the last one and received it today and it appears to be DOA. Can't see it in BIOS, Linux (dmesg), Windows (diskmgmt.msc) or via a USB to SATA adapter. Tried multiple PCs and even taped the 3.3V pin for giggles. Nothing.
The drive I received doesn't match the serial number of the auction/CrystalDisk Info screenshots, so I contacted the seller about a return/refund.
Edit: After trying the procedure mentioned below by luckylinux (leaving the drive connected to power but not data for some time) it is recognized. Very bizarre, but seller is legit.
Last edited: