Lucid Nonsense


DroboPro

Thursday, 23 April 2009

I saw the release of the DroboPro a few weeks ago, thought it was neat and then promptly forgot about it. I was asked about it the other day though so thought I’d put some thoughts down.

The original Drobo was a bit of a revelation when it was released a few years ago because it brought a lot of previously high-end RAID features down from the “thousands of pounds” level into the realm of “a few hundred”. It gives a good level of resilience to hard drive failures, the most flexible upgrade path of any RAID-like device I’ve come across and a pretty straightforward and easy to use interface. If you’ve not come across a Drobo before they have an informative, if a little cheesy, demo video that covers the basics.

The DroboPro’s main feature over the standard Drobo is the doubling of the original’s drive complement, from four to eight SATA drive bays. As with the original you can mix and match different size drives, with a current maximum amount of raw storage, using 2TB SATA drives, of 16TB. You would lose 2TB of that to the data provisioning/protection system (BeyondRAID) that the Drobo employs, but that’s still 14TB of data as a maximum, and when 3TB SATA drives hit the market that maximum will jump to 21TB. Of course 2TB SATA drives are currently very expensive, around £220 ex VAT at the moment, so a £990 DroboPro with 14TB of available storage would be around £2700. A more common option at the moment is likely to be the use of 1TB drives, which are literally a third the cost of 2TB units. At that level a DroboPro with 7TB of available storage would cost a very reasonable £1500 or thereabouts. That’s a respectable increase over the Drobo, as without going to the really pricey drives the maximum using 1TB drives in the standard Drobo is 3TB, not a small amount of storage but even a small design studio can quickly fill 3TB these days. This is especially so if they’re also using the Drobo for Time Machine backups.

Having eight drives available does bring up the possibility of more than one drive dying at once, or a second drive failing shortly after the first. There’s not a huge risk of this happening, but in a Drobo (or a traditional RAID5 unit) that would mean the total loss of your data, and the odds of that happening are actually a littler higher than you would like to think. Acknowledging that higher risk due to the greater number of drives, Drobo have added a “dual disk redundancy” option to the DroboPro. Using that you lose another drive’s worth of available storage but can cope with two drive failures before losing any data. It’s a reassuring option for the pessimistic data hoarder. A slightly surprising but welcome addition to the DroboPro is iSCSI support and Drobo have also included an iSCSI initiator for Mac users; Apple rather unbelievably still haven’t added native iSCSI support to OS X. iSCSI when used properly should beat even Firewire800 in the speed stakes, but we’ll have to wait and see whether the DroboPro can actually feed the interface data quickly enough for that to be noticeable.

As with anything there are a few things to be aware of with the DroboPro, nothing deal-breaking but things that need to be taken into account. Firstly, if it’s anything like the original Drobo, it won’t be that quick. Not unbearably slow and absolutely fine for mass storage, fileserving or low-volume mail/web serving, but certainly not up there if you want to use it as direct storage for audio or video editing. I’m hoping to get a unit in to test fairly shortly so should have a better idea of the speed. Secondly, as with the standard Drobo, it needs to be connected to a host Mac or PC, or the separate DroboShare unit, to share files; it can’t just be plugged directly into a network. To be honest I don’t think that’s much of a bad thing for Mac users, AFP support in most NAS devices is, in general, not good and a Drobo coupled with a Mac mini has been one of our preferred “small studio” network storage systems for the past year or so.

There is a further point that is worth highlighting: however good the DroboPro is, it’s not going to be equivalent to a full business-class RAID system. When compared to a unit like the Promise VTrack RAID, that Apple currently sell in place of their retired Xserve RAID units, it doesn’t really match up. It has no dual power supplies, no redundant controllers, no dual battery backed cache modules, no fibre channel, and the VTrack will be several times faster. There is a rack-mount option for the DroboPro but that doesn’t make it “server-level” hardware. It isn’t a replacement for an old Xserve RAID, or an alternative to a VTrack but then it costs a fifth, or less, of the price of those units with similar storage. It’s targeted at a completely different level to those products and their corresponding price points but impressively manages to bring a good selection of features from those units to a price where many people can justify the expense.

In the end the DroboPro appears to be a natural progression from the standard Drobo. It’s not comparable to enterprise-class RAID units, but then it’s not competing with them. It effectively just provides a second tier to the Drobo range; if the Drobo doesn’t provide enough storage for a customer, we can move them up to the DroboPro instead of speccing multiple Drobo units, which isn’t ideal, or stepping up the several thousands pounds to a more traditional RAID system. It’s a useful option to have and fills the gap where someone needs a large amount of storage but not necessarily all the enterprise-level features.


Important Note - If you’re using planning to use Time Machine with a Drobo there are a few gotchas mainly related to the fact that you need to be aware that the drobo lies to you (in a nice way) about the storage size. The safest way to do this is to use a sparseimage to contain your backups, something that backups made via an Airport Extreme or TimeCapsule already do. I’ll stick some instructions up for getting that working on non-network backups in the next day or so.


Previous Entry: "rsync Binary Installer"

Next Entry: "42"