-
Important Announcement for The TrueNAS Community.
The TrueNAS Community has now been moved. This forum will now become READ-ONLY for historical purposes. Please feel free to join us on the new TrueNAS Community Forums
What number of drives are allowed in a RAIDZ config?
- Thread starter Tekkie
- Start date
Hi All,
Interesting point that no-one seems to have covered yet (regarding 2 x failures in rapid succession), Think that these drives are precision manufactured to high standards - This can almost be their undoing.
Imagine an array of 12x (or greater) drives, all bought from the supplier at the same time and commissioned together. All were built to the same manufacturing tolerances using the same materials. All have had the same run time. All have suffered the same power spikes/surges/blackouts/brownouts.
One fails because it's basically come to the end of it's life (worn out) - that was probably the weakling in the pack (the runt of the litter), but they've all seen the same service so potentially they could ALL fail within a few hours of each other (an increasing problem as precision and quality become better and better). This is also because when one fails there is an increased load on the remaining drives, causing a higher stress level etc.... - you all know the rest. This is why it's REALLY important to keep an eye on drive health and have a spare ready to swap in the moment the first fails, and probably have a second on standby just in case, and re-order on next day delivery after the first goes etc...
Because of this I made a point of spacing my drive purchases with a month or more between each (the old NAS allowed ad-hoc addition of drives, until it decided it couldn't add the final drive because that pushed it over the 16TiB drive pool limit by a mere 70GiB! - you know who you are!!) It also means there's an increased likelihood of getting drives from different batches with slightly different tolerances, varying the failure times considerably.
OK, I might be really unlucky and find that this means that 3 fail simultaneously, but the odds on that are extremely small.
Rototype
Interesting point that no-one seems to have covered yet (regarding 2 x failures in rapid succession), Think that these drives are precision manufactured to high standards - This can almost be their undoing.
Imagine an array of 12x (or greater) drives, all bought from the supplier at the same time and commissioned together. All were built to the same manufacturing tolerances using the same materials. All have had the same run time. All have suffered the same power spikes/surges/blackouts/brownouts.
One fails because it's basically come to the end of it's life (worn out) - that was probably the weakling in the pack (the runt of the litter), but they've all seen the same service so potentially they could ALL fail within a few hours of each other (an increasing problem as precision and quality become better and better). This is also because when one fails there is an increased load on the remaining drives, causing a higher stress level etc.... - you all know the rest. This is why it's REALLY important to keep an eye on drive health and have a spare ready to swap in the moment the first fails, and probably have a second on standby just in case, and re-order on next day delivery after the first goes etc...
Because of this I made a point of spacing my drive purchases with a month or more between each (the old NAS allowed ad-hoc addition of drives, until it decided it couldn't add the final drive because that pushed it over the 16TiB drive pool limit by a mere 70GiB! - you know who you are!!) It also means there's an increased likelihood of getting drives from different batches with slightly different tolerances, varying the failure times considerably.
OK, I might be really unlucky and find that this means that 3 fail simultaneously, but the odds on that are extremely small.
Rototype
Following the above post, there is an argument for using the cheaper, consumer level SATA drives. These generally aren't manufactured to the tolerances of server drives (to keep the cost down) and the odds of buying 12X of the same batch are much lower (from the point of someone setting up a FreeNAS system).
Yes, they'll fail more often, but their more sporadic failure rate means that it'll be much more manageable to replace a failed drive when it goes, with the thought that it's far less likely that another is going to die in the time it takes to replace and re-sync the array.
One for all to think about.
Rototype.
Yes, they'll fail more often, but their more sporadic failure rate means that it'll be much more manageable to replace a failed drive when it goes, with the thought that it's far less likely that another is going to die in the time it takes to replace and re-sync the array.
One for all to think about.
Rototype.
- Joined
- May 26, 2011
- Messages
- 654
@Rototype What a thread necromancer you are :D
Anyway you're right that IF there is a flaw then multiple drives could fail at nearly same time (see my post). Yet spacing the purchases by a month might not mitigate such risk. Disks are manufactured in huge batches and if whole batch has some flaw it will be hundreds of disks. The shop where you are purchasing will have some stock which is getting refreshed by their suppliers. You might get disks from the same batch even after a month. Either because the shop did not sold much of the same disks in the meanwhile or they simply got the stock refreshed from same batch. This is of course more valid for small shops. The big ones like Amazon, Newegg, or even the big country-based ones will most probably have quite rapid flow of common HW like HDDs so the chances are lower but still there.
Yet there is easier way than waiting a month or two... Just ASK them to give you disks from different batches! (and maybe mention that you're building NAS as justification). I did it several times already. I just asked to give me no more than 2 disks from the same batch (as i have Z2). They never had a problem of fulfilling such request for me. They even called me once that my delivery will have ONE day delay as they need to wait for additional disks to be sent from their suppliers so they could satisfy my request. They even apologized for inconvenience and asked if i am OK with such delay (lol... I was feeling bad as i was the one who asked for such extra service for no extra cost yet they apologized huh... ).
Anyway you're right that IF there is a flaw then multiple drives could fail at nearly same time (see my post). Yet spacing the purchases by a month might not mitigate such risk. Disks are manufactured in huge batches and if whole batch has some flaw it will be hundreds of disks. The shop where you are purchasing will have some stock which is getting refreshed by their suppliers. You might get disks from the same batch even after a month. Either because the shop did not sold much of the same disks in the meanwhile or they simply got the stock refreshed from same batch. This is of course more valid for small shops. The big ones like Amazon, Newegg, or even the big country-based ones will most probably have quite rapid flow of common HW like HDDs so the chances are lower but still there.
Yet there is easier way than waiting a month or two... Just ASK them to give you disks from different batches! (and maybe mention that you're building NAS as justification). I did it several times already. I just asked to give me no more than 2 disks from the same batch (as i have Z2). They never had a problem of fulfilling such request for me. They even called me once that my delivery will have ONE day delay as they need to wait for additional disks to be sent from their suppliers so they could satisfy my request. They even apologized for inconvenience and asked if i am OK with such delay (lol... I was feeling bad as i was the one who asked for such extra service for no extra cost yet they apologized huh... ).
Last edited:
- Joined
- Feb 2, 2018
- Messages
- 1,401
Hey Rototype,
This discussion never ends and have been talked about in many thread... No need to bring back a 5 years old thread about that...
About consumer grade HDD, just don't do it. Their firmware has not been programmed for NAS. Here is an example :
One hard drive has a hard time reading back a sector. What will it do ?
The NAS hard drive was programmed for operating with other drives in a redundant structure. So for that one, it does not matter not to be able to read THAT data. Other drives will do it. So after a few tries, it will return an error saying "Sorry, can not read that one!". The NAS will turn to the other drives right away, recover the data, serve the client and then go on to fix the unreadable data on the problematic drive.
The other drive is your regular desktop drive. That one considers it is the only one holding that data. So no matter what, it MUST read it back because if not, the data is lost. So the hard drive will fight over again and again, sometimes for minutes before either reading back the data or giving up. Poor performance, poor management of drive space and more problems with that one.
These desktop drives are not meant to run 24/7, to sustain that heat, that much vibration and more.
You are million time better with actual NAS drives.
This discussion never ends and have been talked about in many thread... No need to bring back a 5 years old thread about that...
About consumer grade HDD, just don't do it. Their firmware has not been programmed for NAS. Here is an example :
One hard drive has a hard time reading back a sector. What will it do ?
The NAS hard drive was programmed for operating with other drives in a redundant structure. So for that one, it does not matter not to be able to read THAT data. Other drives will do it. So after a few tries, it will return an error saying "Sorry, can not read that one!". The NAS will turn to the other drives right away, recover the data, serve the client and then go on to fix the unreadable data on the problematic drive.
The other drive is your regular desktop drive. That one considers it is the only one holding that data. So no matter what, it MUST read it back because if not, the data is lost. So the hard drive will fight over again and again, sometimes for minutes before either reading back the data or giving up. Poor performance, poor management of drive space and more problems with that one.
These desktop drives are not meant to run 24/7, to sustain that heat, that much vibration and more.
You are million time better with actual NAS drives.
