- A mirror made of SSDs from different vendors should provide good protection against firmware bugs like that.Samsung PM863/PM863a and PM883 data miscompare issue - Lenovo Server - Lenovo Support DEdatacentersupport.lenovo.com
SSDs with plp?
- Thread starter edge
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
Different vendors in a RAID mirror is good, because chances are that devices will not fail at the same time because of same make or even batch. But not good enough imo. What if the faulty SSD returns corrupted data? Then you have data in garbage out. The OS will not notice and pass on the garbage to the application. That latter one will perhaps, and crash or error out.
If you run Linux and are a data correctness nut, and who here on STH isn't, I suggest a filesystem that checksums metadata and data. Like ZFS (my favorite) or the newer BTRFS with XXHASH or stronger for algorithm. Check out https://people.freebsd.org/~asomers/fletcher.pdf for some of the insane brain power that has gone into designing a system, that can deal with all sorts of "lies" a disk or SSD might spit out.
I usually only pay half of what the usual SSD price might be, if a device will have no firmware updates. Because they all have bugs. HGST SAS SSDs look good otherwise, but are often (?) OEM models that need updates from that specific OEM. Also no power and performance issues because heatsink is well dimensioned as opposed to m.2 modules which only have a copper-sheet below the label.
If you run Linux and are a data correctness nut, and who here on STH isn't, I suggest a filesystem that checksums metadata and data. Like ZFS (my favorite) or the newer BTRFS with XXHASH or stronger for algorithm. Check out https://people.freebsd.org/~asomers/fletcher.pdf for some of the insane brain power that has gone into designing a system, that can deal with all sorts of "lies" a disk or SSD might spit out.
I usually only pay half of what the usual SSD price might be, if a device will have no firmware updates. Because they all have bugs. HGST SAS SSDs look good otherwise, but are often (?) OEM models that need updates from that specific OEM. Also no power and performance issues because heatsink is well dimensioned as opposed to m.2 modules which only have a copper-sheet below the label.
What if the cpu returns corrupt data? (pentium bug, stuff facebook found in modern xeons: Mitigating the effects of silent data corruption at scale)
Or the nic? (red hat fixed a problem with intel nics in 2018 that corrupted data)
Interesting and tough problems indeed, @i386. I think confidence in the CPU+RAM+mainboard can be raised significantly with 96h of Prime95 torture testing. Not sure if Facebook is doing basically the same on an ongoing basis from the article you posted? ECC-RAM seems not optional. DDR5 with its "ECC-lite" built-in will have to be seen. No overclocking. Quality PSU with 5+yrs warranty, showing reasonably trusted component selection by OEM.
And then you will find out that Intel introduces regressions in microcode updates for security flaws of their silicon. Been bitten more than once. So never upgrade that, or at least not without running the MC update on a second identical system for a bunch of months first. Also buy late-stepping CPUs and not the B0 release version, and never the first batch from a new process.
With the unmitigated disaster that is the Intel I219 ethernet solution, and more recently bugs shown in 2.5G chips from them, I lost complete faith in Intel networking chips. If it just has work, I turn to Mellanox or Chelsio.
And before I forget, always keep your 3-2-1 backups current and test restore occasionally.
And then you will find out that Intel introduces regressions in microcode updates for security flaws of their silicon. Been bitten more than once. So never upgrade that, or at least not without running the MC update on a second identical system for a bunch of months first. Also buy late-stepping CPUs and not the B0 release version, and never the first batch from a new process.
With the unmitigated disaster that is the Intel I219 ethernet solution, and more recently bugs shown in 2.5G chips from them, I lost complete faith in Intel networking chips. If it just has work, I turn to Mellanox or Chelsio.
And before I forget, always keep your 3-2-1 backups current and test restore occasionally.
Here are my $0.02. I've used a lot of Intel and Samsung PLP SSDs. I now use Micron exclusively, specifically their 5X00 series of SATA drives (they also have M.2 SATA and NVMe drives). They all have PLP, the Pro and MAX lineup have very high endurance and you can update their firmware through a micron utility you can download for free. They are inexpensive on eBay and I've yet to have a failure. Overprovisioning a MAX drive which is already heavily overprovisioned from the factory gives extremely high steady state performance and you will be hard pressed to wear them out no matter what you are throwing at them.
But their support sucks, their firmware and tools sucks and their 5300 Pro (SATA with PLP, a so-called RAID and enterprise target drive) doesn't support SCTERC/TLER, which is a big blinking neon-sign saying STAY AWAY
And if you open up the drive it just look cheap build, even the solder-joins of the non-surface-mounted capacitors, e.g. compared to Seagate Nytro 1551 which kill it in any parameters (power-consumption, TLER support, performance, build quality etc), unfortunately they are hard to find these days.
Last edited: