Drobo

Initial configuration -- more redundancy?

I just bought the simplest, 4-bay Drobo. Very exciting. I’ve heard some horror stories, but I’ve also heard many success stories. I bought (4) 2TB SATA drives to use with it, and I popped them all in and everything is good so far. With the Drobo I was hoping to get maximum redundancy, so basically 2TB with all four disks mirroring, or maybe 4TB as a RAID1+0. Sadly this is not an option for me and if I slide the initial formatter to 4TB it says it will make two volumes because I will have additional capacity.

So since it wouldn’t let me try 6TB, I had to go 8TB. Under this, I have 5.44TB of capacity and 1.83TB “for protection”. I don’t know how it does that, but I’ll live on a prayer here. Also, the volume says it has over 8TB available, but the dashboard software clearly says 5.44TB. Not sure about that one.

But my biggest question is: can I configure this thing to do MORE redundancy? I see the Pro has a Dual Disk Redundancy option, but I don’t think the standard Drobo does. (dunno why, it’s clearly a software feature)

the volume sizes have no relation to the actual space - its called thin provisioning

you could have a 16TB volume - you can still never put more data on there than you have actual space on your disks

a standard drobo only supports a single drives worth of redundancy - its equivalent to raid 5 (with 3 or 4 disks) - it isnt configurable in any way

If you are on a Mac there’s no reason not to go straight to 16TB. As Docchris says it won’t let you put more on there than you have space for, and when you want to put bigger drives in then the available space will expand without creating new volumes.

The best way to increase reliability is to break in your drives before you install them in the Drobo (if they are new and especially if you buy a set of 3-4 as you did).

I just bought 3x2TB WD GP drives. I run them hard for 12-24 hours (preferably more like 24 hours) before relying on them. I run the WD diagnostics (good for about 7 hours), and then an HDTune surface scan and then I fill them with data and do another WD test and/or surface scan. And then I put them in the Drobo unless they are destined for some other job.

One of the 3 I bought failed about 4-8 hours into testing, during the WD Diagnostics extended test. It passed a full Win7 format and a partial HDTune scan, I think. Reading Newegg reviews, it is very common for drives to be DOA right out of the box and maybe much of that is shipping damage. And then a fair number of failures within about 1-2 months in. I think typically people buy these drives and very slowly fill them with data, such that the drives might not see as much activity the first month or two as I give them the first day or two during my burn in.

If you believe the reviews then it is not at all inconceivable to get 2 out of 4 drives failing within a short period of time, as my one drive failed after only 4-8 hours of hard work. When a fresh drive is popped right into a Drobo and then a few hundred gigabytes written to it, if one drive fails then the Drobo is going into a multi-hour relay that is very tough on the drives because it is not sequential reading/writing like these diagnostic tests but more like continuous random access. And that is why I believe multi-drive Drobo failures are more common than we would like them to be.

There is no magic number as to how much burn in a drive should get. The above is just what I decided to do.

im finding about 20% failure on the 3tb wd’s at the moment (2 from my first batch of 10 - 2 from my second batch of 10)

all within the first week of use

Chris, did you break your drives in? I’m curious how much exercise they got before they failed.

If’s hard to believe WD is getting 20% of their output back but that’s fairly consistent with reviews.

I’ve also noticed that 2.5" notebook drives tend to have much lower reported DOA and early failure rates, at least the decent models like the WD Blacks, which I’ve used for a few years in my own laptop.

nope, just shoved them in (one at a time)

they were ni a raid 6 array so i dont mind too much if one dies during a rebuild