My NAS is far too noisy to be even remotely within range of a USB DAC!
It’s in my man cave, connected by 40 metres of fibre optic.
I was planning to get a new 4-bay Synology NAS next year for my Roon machine. My 918+ will become a backup machine.
What’s the current thinking on the best way to RAID the disks, given 4 bays?
I’m very much a fan of RAID10. Synology Hybrid RAID (SHR) and RAID 5 & 6 have their issues. Processor overhead and potential loss of array due to Unrecoverable Read Errors during rebuild after a drive failure.
I lost an entire 4-bay NAS during rebuild after a drive failure & replacement a few years ago.
SHR2 is very reliable. Never do RAID 5 or 6 on btrfs because there are documented issues with that combo on any OS.
I should have gone into more depth in my earlier reply.
It depends on what you are trying to achieve with RAID or SHR/2.
SHR gives you single drive fault tolerance at the expense of 1 drive’s capacity. SHR 2 gives two drive fault tolerance, but at the expense of 2 drives’ capacity.
For a 4-drive array, SHR’s overhead is 25%, with SHR2, it’s 50%. There’s also a system overhead.
SHR2 over SHR doesn’t really make sense for a 4-drive NAS, or for any home NAS, really.
Here’s a more in-depth discussion:
https://www.reddit.com/r/synology/s/UsO3R5Jj2V
RAID should never be considered a means of protecting data - the only way to do that is by having robust backup arrangements on and off site.
It’s a choice based on availability/performance and capacity overhead.
Mine’s a 12-bay machine, with all HDDs. RAID10 was my choice for performance. ~6 X read and write speed of a single drive at the penalty of a 50% capacity overhead.
Technically, it’s only single-drive fault tolerant, however, the greater the number of drives, the less likely it is that two drive failures will occur on the same mirrored pair.
At the time I got mine, SSD costs were still stratospheric, so 6 HDDs was a low cost for the performance benefits.
I lost an entire 4- bay NAS (DS414) SHR array a few years ago. I had a bad batch of drives. One drive failed and during the lengthy rebuild process, a second drive failed and the array was toast. SHR2 might have saved it, but there was also a very high likelihood a third drive would have also failed.
I had to start again from scratch from the backup.
SHR/2 saves you the inconvenience of having to restore from backup if you lose a drive. In reality SHR2 doesn’t really offer any greater protection than SHR and it comes at a greater capacity overhead (see the above Reddit)
SHR does have certain advantages over conventional RAID, in that you can upgrade drive capacities to increase the overall array capacity or add drives to it if slots are available (or via an expansion bay) without needing a full array rebuild:
Whichever route you go, a robust backup strategy is crucial.
@Graeme_Finlayson - As always, thank you again for your always-thoughtful input.
The only thing I want to highlight is a counterpoint to the statement I’ve quoted above…the benefit of two-drive redundancy is something I deliberately call out in my setup: it allows for greater data resiliency against insidious behind-the-scenes miswrites/corruption that can be mitigated through periodic data scrubbing. When there are multiple copies of data available, it is relatively trivial for the data scrubbing process to identify which data is corrupted and fix it (with what would generally be 100% assurance) to its intended state.
Through this mechanism, and this alone, I consider multiple drives of redundancy a long-term insurance policy for the integrity of my bit-perfect digital music collection.
OOOPS! I just realized that I forgot to mention that I use John Van Sickle’s FFmpeg library in my setup; added in my OP.
It is all about the use case. I ran Roon on I5 and I7 Gen 10 NUCs (small library some 25000 tracks, all about 25% hi-res) and then moved to a pimped 720+ with 16 GB RAM, 2 SSDs as cache and another SSD (esata) for the database.
I can run all my 7 endpoints with CPU utilization between 10-15% @ 192/24. Unless you go into DSD conversion with a convolution filter, that system run perfectly, quick responses and w/o any problem.
For me, it is more than enough for the next few years.
Hi,
i’m using 923+ 32 GB. With four Wd red disks and two synology 400 gb ssd.
Roon and DB installed on ssd. (volume 3)
Very large Music collection on volume 1 two disks.
Very large movie/series collection with Emby on volume 2 two disks.
Everything wired! works flawless
I may have waited too long. My 918+ seems to be dead. But it’s only the power supply brick (the LED isn’t lighting up).
Hopefully it’ll be fine with a new PSU. My Rackstation’s SMPS blew up a few weeks ago. Tripped the main supply GFCI (RCBO) to my man cave.
A new SMPS a couple of days later and it was working perfectly again. No errors, corruptions or any other issues.
I have a Fritzbox 6660 which supports 2,5 gbit/s LAN connection (and my QNAP NAS has a 2,5 gbit/s port too). So using that would be recommended?
pls, check the internal power cell
Sounds like a fine combination.
The internal power cell? In the main chassis, or in the power supply brick? What does it look like?
Are there any improvements to be expected vs both on a 1 gbit/s port? Less jitter etc?
The power cell on the main board of the DS918
There’s quite a Reddit thread about this:
https://www.reddit.com/r/synology/comments/q59ue4/ds918_power_supply_died_what_to_check_when/
Markus,
The advantage of your 2.5 Gbps link (or my LACP 2 Gbps link) versus 1 Gbps is simply more bandwidth to read and write data from/to your storage, which includes multiple users in your house listening to music over independent connections. There is no other benefit.
Perhaps a variant of dollar cost averaging would be helpful here. Buy only one drive a month, each from a different source, over a period of N months, to avoid the possibility of multiple drives coming from a bad manufacturing batch.