Planning for future replacement of Drobo 5D
I'm currently using a Drobo 5D for Time Machine backups. From what I've read, direct attached storage is more reliable for time machine (NAS including the old Time Capsule's apparently back up to disk images, so are more prone to corruption - it sounds like Synology users frequently have to erase their time machine backups and start again).
I'm currently using two Thunderbay 4's with a 6 drive (4TB each) RAID 1+0 for my data (12TB) volume, and backing up to an 18TB Time Machine on the Drobo 5D.
In planning ahead for when the 5D fails or no longer gets software updates (primarily working out what drive sizes to buy when I need to get replacements), would it be a good idea to plan to have a 4 drive RAID 5 for Time Machine backups? eg. 4 x 12TB drives would give me 36TB capacity, or to make use of some of my older drives I could get a 6 bay Thunderbay with 6TB drives for 30TB capacity. I don't need as robust redundancy for a Time Machine backup as I do for my main data volume, and don't need fast write speeds.
Or is there a better solution? Is it safe to rely on SoftRAID for both my main data volume and my Time Machine backups, or would I be better off using different 'technologies'? A Synology NAD seemed like the easiest solution, but if it's less reliable for Time Machine backups over a network it puts me off.
SoftRAID is a great solution. It has been around for 25 years, and the biggest challenge for us is keeping up with the changes in macOS security, which has caused some pain for users, but we are nearly through that gate.
The main difference you will notice is All SoftRAID volume partitions are the same size, so we do not support the mix and match in Drobo, which I am not convinced is reliable.
You get feedback on the health of disks and if the 30TB is adequate for you, then go with the less expensive smaller drives!
@softraid-support I'm also looking into potential replacements for a Drobo. Looking into Thunderbay and SoftRAID, and had some additional questions (hope it's OK to combine it into this thread instead of making a new one):
1) I've tried to read a bit about the mix and match difference - it seems you can mix drives of different sizes, but the volume size will only act as if all of the drives are the same as the smallest drive. Is that right? So you can still upsize without needing to create a new RAID volume (although this is reportedly what most SoftRAID users do) - you can still do the Drobo way of removing one drive, adding a new one, allowing the RAID to rebuild, and then replacing the next drive. SoftRAID would just involve an added step of forcing a resize once you were done?
2) I've read that the RAID size, based on number of physical disks, is set from the beginning. So if I wanted to go from something like a Thunderbay 4 to a Thunderbay 8 in the future, I could not just add in some extra disks and then add them into the array, I'd need to completely redo the array. Is that right?
3) Does redoing the array mean that pre-existing data is wiped and the new array starts with zero data in it?
4) I noted the main SoftRAID page prominently displays that version 6 of the software is in public beta, implying that it's coming soon. Does the company plan to do something like offering a free upgrade to version 6 for anyone who has purchased a new license in the 6-12 months before the new version is released?
Thanks for your time!
1. Yes. But also, SoftRAID does not do anything to the file system that is non standard. A SoftRAID volume is the same as an Apple volume, its the wrapper around it that is different.
2. Yes, you have to delete/recreate the volume.
3. yes, necessarily.
4. There will be an upgrade program, but details have not been finalized, nor announced at this time. I would expect Big Sur to be about the break off point, but that is not my department.
@softraid-support thank you for the reply, I appreciate it. Do you have a sense for how users might handle physical drive expansion? For example, if I want to go from my five-drive Drobo to a Thunderbay 8, it sounds like it wouldn't be possible to load the Thunderbay with a few drives, copy the data over, then put in some of the Drobo drives, wipe them, and integrate them into the Thunderbay-based array. In that situation, the only option is to fill the Thunderbay from the start. Is that right?
If I do have it right, then one more question: I looked over the page for the upcoming SoftRAID version 6 and didn't see a mention of being able to change the number of drives in the array without wiping on the product page, either. So I assume it's not a soon-to-come feature, but again just wanted to check to see if it might be.
Thanks again for your time!
There is no way to move a Drobo "one disk at a time" It is a form of RAID, so you need all disks (minus 1) to mount.
So you need to copy all the data over first.
We do not have immediate plans to expand a RAID volume by adding disks. Maybe someday, but we have no current plans
@softraid-support Sorry, to clarify: if I wanted to go from the Drobo 5-bay model to the Drobo 8-bay model, I could theoretically fill the 8-bay model with three or four disks to build an initial RAID, copy the data from the 5-bay model, and then add in some of the disks from the 5-bay model. The system would integrate the disks into the RAID and expand it, without needing to wipe the active RAID. (Actually, if I wanted to upgrade and keep within the Drobo line of products I could just move the original five drives over and then add 1-3 new disks, but I used the example because that's what I'd need to do with switching over to the Thunderbay + SoftRAID system.) In that way you can recycle some of the disks, instead of needing to have your new RAID array completely built.
For me, it's good to know going in that I'd need to have the RAID totally set and ready to go with its own disks. If there's a way to submit feature requests, being able to work with disks of different sizes, and adding in new disks without needing to wipe the array and start from scratch would be some features I'd love to see. For example, for now I'd opt for the 4-bay Thunderbay but in a few years I could see myself needing and upgrading to the 8-bay Thunderbay. Knowing I'd need to have all of the disks for each from the start (and that the remaining ones would essentially become redundant once the RAID was replaced) seems less flexible
Thanks for the ideas, but we are unlikely to ever support variable stripe unit sizes (different disk sizes), as it is a mess. It requires fooling the file system in addition.
Possibly we may add increasing the disks in a RAID, but even that is outside our scope. When you do not control the "hardware" (as in hardware RAID) moving data around is complex and fraught with places things can go wrong. So it is unlikely we will go there.