Another vote here for Synology (can’t beat them for price and reliability) and more, smaller, drives.

For weeks I've been weighing the Synology vs QNAP decision. QNAP seems to have more sophisticated hardware, but the interface is more important to me with this project.

I appreciate knowing the WD Red drives will be the best choice.

quote:

Originally posted by PeterGV:
Another vote here for Synology (can’t beat them for price and reliability) and more, smaller, drives.

QNAP is just as good if not better.

quote:

Originally posted by JimTheo:
If I were you, I'd be considering a raid 5 with warm spare solution as any single failure will not be catastrophic. Expandable if needed is a perk.

What is a warm spare solution? With Raid 1 I used to pull a drive on a regular basis, put it in storage, and rebuild the array. How would the backup process work with Raid 5?

I ask this because having chatted for a while yesterday with a tech rep for the NAS manufacturer I'm considering, she suggested an array which has 5 drives, setup as Raid 5.

What might the pluses and minuses be for a five drive Raid 5 NAS? This solution would be more expensive on the front end, but allow for expansion later.

There is no redundancy with a RAID 5 array.

There is a fault tolerance of 1 drive, meaning one drive can fail without affecting the array (two drives failing will cause data loss), but this is not the same as redundancy.

You will still need to keep another drive attached, or backup to the cloud, for true redundancy. Many NAS units have USB ports so that you can hook up cheap mass storage to create automatic redundant backups.

You can also setup a 4 drive unit into a 1+0 RAID if you want redundancy, but now you are looking at four 6 TB drives to get 12TB of space.

Personally, I would just go with four 4TB drives setup with Synology Hybrid Raid and keep a spear matching drive around to swap in should one ever get weird.

Aeteocles, your points are helpful to the discussion. If I understand this correctly, there is no redundancy advantage to having 5 disks over having 2 disks. There would be an advantage to using an 8 disk array in a 1+0 RAID.

I talked this over yesterday with the tech rep, and oddly enough there may be some advantage in my situation to start with a two disk array, planning later to add another two disk array, in a RAID 0+1 if I plan to regularly swap out drives.

That is really only if I insist on using the fireproof, flood proof NAS. Otherwise the choice of a larger array of smaller drives is clear.

From a fault tolerance standpoint, a two drive RAID 1 will behave the same as a RAID 5 of any size. Both will give you a fault tolerance of 1 drive before the array fails.

But that's not the same as redundancy. Redundancy means that there's more than one separate and distinct copy of the data. RAID 1 gives you redundancy, RAID 5 does not.

You have to decide for yourself your risk tolerance. Do you truly want redundancy, or is fault tolerance just enough? Does having a second copy of your data provide a benefit for you? Are you using the second copy to protect against malware our ransomware? Is your second copy going to burn in the same fire or down in the same flood that takes out your first copy?

My keep one copy of my data in the cloud. That's my redundancy. Therefore, my RAID array is setup just for fault tolerance.

Redundant
Array (of)
Independent
Disks

quote:

Redundancy means that there's more than one separate and distinct copy of the data. RAID 1 gives you redundancy, RAID 5 does not.

I disagree with your semantics and your conclusion.

RAID merely protects from drive failure so it is wise to also backup the data.
The more places the better.

Call it what you like.

How does a RAID 0 protect against drive failure?

quote:

Originally posted by Aeteocles:
How does a RAID 0 protect against drive failure?

It doesn't.
Ironic huh?

quote:

Originally posted by smschulz:

quote:

Originally posted by Aeteocles:
How does a RAID 0 protect against drive failure?

It doesn't.
Ironic huh?

Theoretically it doubles your risk of data loss. The only point is speed.

Skip RAID 5, IMO. It was only ever useful when disk was expensive. For an array of X drives you'd lose only 20% to redundancy. Add a hot spare in a system that would automatically swap it in, when needed, and you could suffer a double drive failure w/o data loss.

But RAID 5 has a couple of corner-case failure modes that can result in total data loss (don't recall them atm), so, with disk prices what they are, now, I'd only go RAID 10. Or simply RAID 1, if you're using SSDs.

quote:

Skip RAID 5, IMO. It was only ever useful when disk was expensive.

Drives still can get expensive especially the larger, better quality ones.

Choosing a RAID level and even whether to use RAID at all can vary greatly depending on circumstances.
Even using RAID in servers today the philosophy is changing.
Especially with flash (SSD) storage, Windows Storage Spaces (Server 2012/16) coming into prominence.
Additionally there are whole host of charges associated with deploying RAID other than just the disk cost > enclosures (hot swap backplanes), the controller card itself and the number of drives needed.

THEN you need an equal if not greater solution for data backup.
It can add up fast.

Even for desktops RAID is pretty much unnecessary and performance is generally more of a concern especially with the advanced storage technologies of SSD, m2/PCIE, Optane Memory and faster processing.


It is something you need to think through as opposed just accepting that RAID is the only answer.

When dealing with Raid 1, what is a good practice for how much of the capacity should remain empty?

Holy hell, how long would it take to complete a raid 1 12tb rebuild with 7200 rpm sata?

At what risk to the surviving drive? Doing a rebuild stresses the surviving disk.

I don’t quite understand your question about reserve space on a R1. This data is deep storage, right?

I’m not a fan of huge sata drives. Man I wish sata would die. More spindles, smaller drives, especially if you’re going to need to access with any frequency.

quote:

Originally posted by deepocean:
When dealing with Raid 1, what is a good practice for how much of the capacity should remain empty?

What do you mean by 'remain empty'?

1. How much is used by "overhead"?
2. How much of the allocated space should you use on the drive?

1. RAID Usage Calculator
2. You can use pretty much all of the allocated space.

quote:

Originally posted by smschulz:

What do you mean by 'remain empty'?

How much of the allocated space should you use on the drive? <----this

quote:

Originally posted by smschulz:

1. RAID Usage Calculator
2. You can use pretty much all of the allocated space.

Thank you. I remember in the past it was said by some to leave a certain proportion of the drive's capacity unused (my technical term is empty )

I'm going to start with a simple two drive setup, with 10 TB drives in a Raid configuration, and a few hot spares which I will swap in and out on a regular basis (TBD).

I was going to use a 5 drive array, but I think it better to start simple. When I fill up the 10 TB, or get close, I will reevaluate as needed.

I was unsure whether to go with WD Red or WD Red Pro, but I think for my non-commercial, intermittent archival needs the WD Red is the best option.

I use WD Gold or Red Pro.
Gold for regular server use and Red Pro for backup, NAS or other storage.
Red runs at 5400rpm and GOLD or RED Pro at 7200 with more cache.
YMMV

If I use Red Pro, will the raid rebuild faster?

quote:

Originally posted by deepocean:
If I use Red Pro, will the raid rebuild faster?

Maybe a little but the rebuild rate is more a factor on the RAID card.

quote:

Originally posted by smschulz:

quote:

Originally posted by deepocean:
If I use Red Pro, will the raid rebuild faster?

Maybe a little but the rebuild rate is more a factor on the RAID card.

Is there any negative in terms of reliability/failure rate related to the increase in rpm?