Hello again,
I am still debating about a reasonable way to go for my little home NAS. Previously I had posted the question about a fanless NAS, which I have understood and concluded as not the way to go for a NAS using spinning disks.
As this is the first time for me to set up a NAS, I am still figuring out which way to go, and would be grateful for any feedback.
I have quantified my needs:
Option 2 could grow more in sync with my needs.
I have two questions, basically:
The most economic one is the 2+1 option, closely followed by the 1+1.
The 4+2 options are more expensive, but close to each other. The more expensive 2.5" format is (to this date) being more than overcompensated by energy savings (3 years runtime).
Prices for replacement disks are not included.
So what about failure rates? Regarding The Math on Hard Drive Failure and adding the missing 1+1 case, these are the numbers:
The 0.876% corresponds to the specification of MTBF=1*10^6 hours. The other disk rates are within the range of observation from the Google publication, essentially copied from the above mentioned thread.
It appears that with increasing AFR (aging disks) the 1+1 solution performs best!
However, I find AFR numbers not too insightful. The array AFR represents the case that the array would fail in one year if completely unattended (as I understand the underlying math). That means that a maintained array (someone replacing a failed disk quickly) would be less risky than these numbers show.
Can you compare your experience to these figures?
Thanks a lot!
I am still debating about a reasonable way to go for my little home NAS. Previously I had posted the question about a fanless NAS, which I have understood and concluded as not the way to go for a NAS using spinning disks.
As this is the first time for me to set up a NAS, I am still figuring out which way to go, and would be grateful for any feedback.
I have quantified my needs:
- Right now...
- I have about 1.2 TB of data,
- and had about 3 TB if I would do everything I want at the moment.
- In 3 years from now...
- I might have 4 TB realistically,
- and would have about 8 TB if I started all the projects I would love to.
- going with one big vdev for the next years (8 - 9 TB now),
- starting with one vdev now (4 TB), adding a second vdev in - say - 2 years with about 6 TB, and then continue every 2 years or so in a tic-toc-manner replacing the respectively older vdev.
Option 2 could grow more in sync with my needs.
I have two questions, basically:
- Are there real-life issues with the above strategies? Could you contribute some experience?
- I am trying to get a "feeling" for it by doing some math. Below, I created some numbers for different HDD arrays for the latter option (4 TB to begin with). Do my estimations make sense in practice?
- base system: 10 W idle, 42 W load, 16 db(A) fan @ load, price 550 €
- average load: 20 %
- 0.27 € per kWh
- 1+1 WD Red 4 TB: system: 23.5 W average, build price 846€, 3 years energy cost 167 €
- 2+1 WD Red 2 TB: system: 26 W average, build price 793 €, 3 years energy cost 185 €
- 4+2 WD Red 1 TB: system: 32.4 W average, build price 910 €, 3 years energy cost 230 €
- 4+2 WD Red 2.5" 1 TB: system: 21 W average, build price 964 €, 3 years energy cost 150 €
The most economic one is the 2+1 option, closely followed by the 1+1.
The 4+2 options are more expensive, but close to each other. The more expensive 2.5" format is (to this date) being more than overcompensated by energy savings (3 years runtime).
Prices for replacement disks are not included.
So what about failure rates? Regarding The Math on Hard Drive Failure and adding the missing 1+1 case, these are the numbers:
The 0.876% corresponds to the specification of MTBF=1*10^6 hours. The other disk rates are within the range of observation from the Google publication, essentially copied from the above mentioned thread.
It appears that with increasing AFR (aging disks) the 1+1 solution performs best!
However, I find AFR numbers not too insightful. The array AFR represents the case that the array would fail in one year if completely unattended (as I understand the underlying math). That means that a maintained array (someone replacing a failed disk quickly) would be less risky than these numbers show.
Can you compare your experience to these figures?
Thanks a lot!
Last edited: