Which HDD's for backup.

EB-1
Offline
• • • • • • •
Upload & Sell: Off

p.2 #1 · p.2 #1 · Which HDD's for backup.

Bruce n Philly wrote:
I use 2 NASs with four drives each... found it was cheaper to add a NAS than upgrade disks... I purchased a used, matching NAS from ebay, and then populated with new drives. My NASs are not exposed to the internet. I have the NASs spin down after one hour of non use to preserve disk life.

Then I backup the NASs to other single drives I connect up when I do my big backups about once a month. I have triple backup redundancy not counting the working drive in my workstation.

I use FreeFileSync... I can pretty much just...Show more →

If you have not had failures you have not used enough devices - that goes for SSDs, HDDs, etc. I've had failures of M.2 Samsung SSD, WD SATA III SSD, Seagate NAS HDD, and even an Intel (now Solidigm) enterprise U.2 drive in the past few years. You are right that data is safer on SSDs since the error rates are two orders of magnitute better for SSDs than the best HDDs (10^-17 vs. 10^-15 UBER), but when they die they die and often suddenly with no SMART warnings.

I am cognizant of the old 321 storage model, so I have at least one local NAS of brand Q and then another of brand S. I would never use two of the same model in case there is some inherent design defect. I also might populate one with Seagate EXOS and one with WD HC5xx drives for example. Offsite storage might be older models/drives.

EBH

Sep 09, 2024 at 11:36 PM

EB-1
Offline
• • • • • • •
Upload & Sell: Off

p.2 #2 · p.2 #2 · Which HDD's for backup.

Chris Dees wrote:
I think I’m looking for somenting like This
Qnap DAS

I have the 8-bay version of that, but it's really just a USB bridge to a simple SATA PM with a wimpy controller that does that weak firmware RAID. They are best used as expansion for a QNAP NAS where the JBOD mode of the enclosure is used with the CPU and OS of host NAS creating the array and powering the parity calculations. As a standalone you could use RAID 1 or RAID 10 I suppose, but there are other options for that also.

I was literally about 5 minutes ways from throwing the box out in the trash because if it does not interact well with the host it will mark a drive bad and there is no easy way to fix it without a 24+ hour test in that enclosure.

EBH

Sep 09, 2024 at 11:49 PM

EB-1
Offline
• • • • • • •
Upload & Sell: Off

p.2 #3 · p.2 #3 · Which HDD's for backup.

Chris Dees wrote:
I would only connect a NAS directly to my MBP dock (just a DAS) and only turn it on when I need it. This seems like an expensive solution to me. A NAS easily costs €400,- and then I still have to fill it up with disks. In addition, something has to be done to make a backup of the NAS.
I have all my photos on an 8Tb Raid0 SSD that is connected to the MBP and is turned on when I need them.

For now I am thinking of the following:
2x 12/14Tb disks in a Raid1 setup (Acasis has...Show more →

I think those last two are probably your best low-cost optiona if the hard drives are good ones.

EBH

Sep 09, 2024 at 11:51 PM

Bruce n Philly
Offline
• • •
Upload & Sell: Off

p.2 #4 · p.2 #4 · Which HDD's for backup.

EB-1 wrote:
If you have not had failures you have not used enough devices - that goes for SSDs, HDDs, etc. I've had failures of M.2 Samsung SSD, WD SATA III SSD, Seagate NAS HDD, and even an Intel (now Solidigm) enterprise U.2 drive in the past few years. You are right that data is safer on SSDs since the error rates are two orders of magnitute better for SSDs than the best HDDs (10^-17 vs. 10^-15 UBER), but when they die they die and often suddenly with no SMART warnings.

I am cognizant of the old 321 storage model, so I...Show more →

Yes, I guess I have been lucky with SSDs and they fail with no warnings. Your comment about SMART warnings for spinners is something everyone should care about but I don't think most have any idea what that is. I won't go into a tutorial here, but Windows and your NAS has features where you can monitor SMART statistics that give you warning of a pending failure. NASs are better at this as they will send you an email if this is triggered and you can then proactively swap out a drive... which I have done. So buying a NAS is the first step, but you must make an effort to configure it to send you emails.

Regarding multiple brands of NASs... yes, you are going the extra mile to mitigate risk. I chose the same model for ease of management as I only need to know one system. Also, my NAS was old and solid... so I assumed (danger here) that it is a good unit and had in its day, great reputation. They were rather cheap on eBay. So far, the system is super solid (as they are until they fail eh?). But I feel good about all of this. And again, as I noted, I have these NASs backed up to two different places so I am super covered.

My experiences with spinner failure is so much that I start out simply not trusting them at all. I use them, but I don't trust them... they will fail and any user of these should plan their system and management practices with this assumption in mind... which is why I use RAID 5 in the NASs... they will just fail. I configured my NASs to spin down the disks after one hour of inactivity which I hope will extend their life.

Peace
Bruce in Philly

Sep 10, 2024 at 07:30 AM

Bruce n Philly
Offline
• • •
Upload & Sell: Off

p.2 #5 · p.2 #5 · Which HDD's for backup.

Another tip: Clone your C drive.

In addition to all of this backup stuff, I clone my C drive to another SSD. I have a snap in slot on the front of my main workstation... Now since all my data, LR catalog etc are on my D drive inside my PC, should I have a C drive failure, I can just snap in the cloned drive and tell the BIOS to boot from there. Buy a new C SSD drive and clone back.

If you are in an environment that requires super-high availability, you can install the clone drive inside the PC and instruct the bios to boot from C, and if not, from your clone drive. No loss of work, however you may not know you had a failure as the machine is just running.

Peace
Bruce in Philly

Sep 10, 2024 at 07:39 AM

Alan321
Offline
• • • • • •
Upload & Sell: Off

p.2 #6 · p.2 #6 · Which HDD's for backup.

I consider a sudden drive failure to be merciful

Gradually worsening reliability is more likely to involve corrupted files over an unknown duration and that complicates things - especially if I'm not using those files until long after they have been corrupted and they have already been incorporated into regular backups. It's partly why I keep multi-generational backups. i.e. I keep what ought to be outdated backups as well as recent backups. Occasionally, I'll cull from them any data that exactly matches current data. Eventually I'll scrap them.

A 4-drive RAID holds a lot of data that really ought to be tested regularly. I use md5 checksums to detect any changes. Doing those calculations on all of those files involves a huge amount of data transfers and for that I want a speedy transfer interface. These days, 5Gbps is well short of being optimal. My LaCie 6big DAS can deliver 500-800 MegaBytes/s of sustained data transfers with sequential data (not flitting about all over the drives for different files or different parts of files). That's with a six-drive RAID 5, two of which are for parity. For read purposes it is like a 4 drive RAID 0.

I happen to keep several volumes on my DAS, usually with not all being "mounted" at once. I can copy one or part of them onto a speedy internal 4TB SSD and then do the testing there. This maximises HDD transfer speeds and testing speeds at the expense of having to read the data twice (once from the DAS and once from the SSD).

I don't and won't use a backup volume for very long. Eventually I park it and start a fresh backup. It's a balancing act. More time to occasionally create and populate a new backup volume vs less time to frequently read all of its data for file verification. Recovery from incremental backups is slower than from snapshot backups, but also potentially more thorough.

Don't forget that the RAID box can itself fail even if the HDDs are ok. If it does fail then you might still lose all data during the failure or even later on if the drives are transferred to another RAID box. It surprised me when I discovered the hard way that a RAID controller may automatically initialise the HDDs. Luckily I had my backups of the backups.

Having written all this, I should add that I am cursed. My failure rate of all things computer, even with high quality gear, is way higher than normal. You are very probably not that unlucky but if that's the case then you may also be less prepared for dealing with it when something goes very wrong; especially if it's a painful gradual process that has been happening for quite a while before you notice it, or if it's a total FUBAR, or if it's intermittent.

- Alan

Sep 10, 2024 at 12:34 PM

Alan321
Offline
• • • • • •
Upload & Sell: Off

p.2 #7 · p.2 #7 · Which HDD's for backup.

Bruce n Philly wrote:
Another tip: Clone your C drive.

Excellent advice, Bruce, and very easily overlooked. I go a step further by having several SATA Windows OS drives that I rotate periodically. Each is labelled with the date of backup. I can swap the current C drive with one of the backups and be up and running in literally a couple of minutes. It was even quicker when they were all SATA drives used in a removable-drive bay, but my primary C: drive is now on an NVMe m.2 card for greater performance between failures. The reason for several copies of C: is that some or even many faults are not initially apparent and by the time I do notice it I may already have done more than one backup.

Macrium Reflect, which I use for most backups, has a live drive-cloning option that will copy my C: onto a SATA SSD in about 8 minutes. It has saved my computer and sanity (well, maybe just the computer) several times. It would take me days or weeks or even months to fully reinstall and configure Windows and everything else on my computer.

Because I use this system, I do not need to include C: in my routine (daily) data backups.

A less significant but still useful option is to manually create Windows Restore Points before and after installing new software or Windows Updates. The before may get Windows working again. The after may save you having to completely reinstall software again.

- Alan

Sep 10, 2024 at 01:00 PM

Alan321
Offline
• • • • • •
Upload & Sell: Off

p.2 #8 · p.2 #8 · Which HDD's for backup.

Some time ago I bought a two-drive RAID case and put two 8TB HDDs in it, configured as RAID 0 (striped, for greater performance).

In retrospect this was a poor decision because I soon realised that ...
1. Such a pair of drives plus RAID controller is at least twice as likely to fail as any single HDD.
2. A RAID 0 is all or nothing when it comes to operational integrity. Any failure = total failure.
3. A RAID 1 (mirrored) performs almost as fast as a single HDD, which is far too slow to impress me or satisfy me since I got used to SSDs - especially for random access work.
4. 16TB is far too much to lose for one failure.

To reinforce this, it turned out that both drives had intermittent faults and were replaced a several months later. If only one was faulty, or if the faults were sustained, then I'd have diagnosed it quicker. So with this RAID, I very quickly lost the warm fuzzy feeling of buying and using good hardware. Instead of being near-primary storage this RAID is mostly just a speedy-ish backup of my DAS.

- Alan

Sep 10, 2024 at 01:20 PM

1 2 end
FM Forums \| Post-processing & Printing \| Join Upload & Sell 1 2 end

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.