What are the benefits of larger arrays?

Currently reading
What are the benefits of larger arrays?

21
0
NAS
DS920+
Operating system
  1. Linux
  2. Windows
Mobile operating system
  1. Android
  2. iOS
Hi,

I was wondering about the various enclosures that Synology has and the upper limit of 108 TB per volume.
Is there any benefit of using the larger arrays like those 8 bays vs a 4 bay?
I can only think about using SHR-2 vs SHR but nothing else beyond that.

Thanks!
 
As both an 8-bay and 4-bay user there are obvious benefits in terms of storage available, file access, network load, IOPS, throughput, redundancy and resilience when compared to a humble 4-bay. My 8-bay has roles and tasks that are simply impossible on a 4-bay unit.

☕
 
I can understand about the storage available assuming that the size of the drives are similar.
But if I were to use higher capacity drives on a smaller unit, I can attempt to compensate for the difference.

I am assuming that network load and IOPS is due to the load being split across more disk and therefore achieve a higher throughput.

But for redundancy and resilience, I am not sure how it comes into the picture.
I understand for SHR and SHR-2 I can lose up to 1 and 2 disk respectively, but that is the same regardless of the number of bays beyond 3 and 4, right?
 
Last edited:
In reverse order:

SHR (RAID 5 and a bit) is a particularly unappealing way of looking-after HDD data. When a HDD dies the remaining drives when a replacement is added are thrashed for a day or more, with no protection, in the hope that you don't kill one of the remaining HDDs in the process or stumble on a bit-flip or other physical error. It is a very bad gamble and RAID 5 was depreciated many years ago and the problems only get worse with larger and larger drives feeding data through the existing (and very slow) letterbox. SHR/RAID 5 does have a use with SSDs though, as the chances of surviving a rebuild is at least 2 orders of magnitude better and the exposure time is much reduced.

Yes, more bays gives more IOPS potential with spinners but, more importantly, it gives you flexibility on how to use your storage and any reserve bays or drives. For example, on my own 8-bay I originally had 5 SSDs (now 6) providing a very high performance volume plus the last 2 bays as a large capacity HDD backup and archived work, which is also mirrored. If/when (and had to do twice recently) I can do a clean restore of the SSD volume with drives on the same backplane, so no load on the network and it is as fast as the drives can muster (I do have another on-prem backup to cover a simultaneous failure of both volumes).

In truth, you cannot compensate for less bays with extended HDD capacity - the 8-bay can equally use such drives to provide the bulk and do other things or reduce/avoid the lengthy and worrisome issue of a RAID rebuild.

The convenience and capability of the larger capacity units is the flexibility to work as you need it to. Of course, never underestimate the power of an empty bay when you need to check something, repair in-place or isolate an issue. It's not always about sequential throughput or mass storage.

☕
 
So if and when possible, I should be looking towards a SHR-2/RAID 6 kind of setup in the long term.
To at least account for the risk of another disk failure during the rebuilds.

Would the rebuild of a 6 disk array take longer than an 8 bay array assuming the total volume is the same?
E.g. 6 x 16 TB vs 8 x 12 TB?
Trying to understand better the whole rebuild process and what factors impact the rebuild speed.
 
Would the rebuild of a 6 disk array take longer than an 8 bay array assuming the total volume is the same?
With DSM7, the rebuild will affect only used space, so the more utilized volume will take more time, and not necessarily the total size of the volume.
 
RAID 10 is probably an ideal thing - mirror of stripes, but the storage cost is high and you won't get maximum storage capacity. IMO SHR-1 with an external backup of your NAS would be optimal. I would not run RAID6/SHR-2 unless this is for a business and concerns about maximum redundancy and high uptime requirement. It is a dog to expand and rebuild and puts stress on the entire array when doing so. I ran SHR-2 on my first DS1815+ AND spanned a storage cabinet, which was a bad idea and wouldn't do either again....
 
How does someone measure IOPS? I am running a couple of SHR-1 arrays on 6 bay devices, with only 1gb LAN, so I am maxing out the network before any read/write performance kicks in, how do you measure IOPS (which is a different variable right?)

I am doing exactly what @Coop777 says above, SHR1 with backup to another nas also running SHR1.
 
Correct me if I am wrong, but for RAID 10 or any form of standard or nested RAID levels, the storage cannot increase dynamically?
What I previously understood was it is defined during creation and even if you swap out for a larger drive as a replacement, it only allocates what what was previously defined with the rest being an unallocated drive?

IMO SHR-1 with an external backup of your NAS would be optimal.

My current NAS is being used for PC backups and photo backups and is running SHR-1.
I don't know if I will do an external backup for it but I was hoping it would be a one stop shop for the whole family since this is my first NAS.
Maybe when I retire my DS920+ for something else, I would repurpose it as a cold backup?
 
RAID 5 / SHR1 HDD arrays are for sacrificial data. The probability of failure is high but this matters little if you have a robust backup in place or when you only use a RAID 5 / SHR1 NAS as a backup-only device (ie with primary data somewhere else but using it for Time Machine, iMazing iOS backups etc).

If you ever build an array and worry that you may have to frag it and start again then you are doing it wrong. Otherwise you get the 'Running With Scissors Award':

Scissors.jpg


☕
 
RAID 0 has no parity; if a drive fails the array is simply dead, instantly. Dead parrot time.

No other niceties either, such as BTRFS data scrubbing as there is nothing to fall back on when a bit is flipped. RAID 0 has its uses but bulk-storing data is not one of them.

☕
 
RAID 5 / SHR1 HDD arrays are for sacrificial data. The probability of failure is high but this matters little if you have a robust backup in place or when you only use a RAID 5 / SHR1 NAS as a backup-only device (ie with primary data somewhere else but using it for Time Machine, iMazing iOS backups etc).

If you ever build an array and worry that you may have to frag it and start again then you are doing it wrong. Otherwise you get the 'Running With Scissors Award':

View attachment 10183

☕
disagree. SHR-1 and RAID 5 have been very stable for me, you need good quality NAS or Enterprise rated drives, of course.
 
disagree. SHR-1 and RAID 5 have been very stable for me, you need good quality NAS or Enterprise rated drives, of course.

I'm glad as it would be a miserable experience for you. Everyone knows a smoker who has never had a health issue but on a broader scale it would be unreasonable to suggest that it does not present serious risk based on that self-selecting group.

RAID 5 really is depreciated and the reasons for that have only increased exponentially since then. You can run RAID 5 / SHR1 in certain circumstances but capacious HDDs with a slow interface and a barely improved mechanical lifetime isn't one of them.

☕
 

Create an account or login to comment

You must be a member in order to leave a comment

Create account

Create an account on our community. It's easy!

Log in

Already have an account? Log in here.

Similar threads

  • Question
Good to hear. You could also clone the drive and expand the data partition using Clonezilla/gParted.
Replies
7
Views
1,102
Thanks - Good thinking on the ps. Both drives happened at the same time - so you may be 100% correct. I...
Replies
2
Views
1,243

Welcome to SynoForum.com!

SynoForum.com is an unofficial Synology forum for NAS owners and enthusiasts.

Registration is free, easy and fast!

Trending threads

Back
Top