Usability: Corporate Price Point vs Consumer Price Point

[thorfinn]

I bought a Drobo, and configured and formatted it last night. Copied the data from my previous external storage disk (which was failing with block errors). Plugged it into the back of the Airport Extreme, and voila, it's just working. Time Machine is happily grinding away doing its thing, and it's very nice knowing that we have 3TB of raid storage that is protected against single disk failure. I didn't have to install any drivers, work out any raid configuration details, or fiddle any settings - there essentially aren't any to fiddle.

This is essentially representative of what I'm enjoying about the current state of play in computer technology - quite a lot of things are falling out of the Corporate Price Point down into the Small Business / High End Consumer Price Point.

A few examples:

• Mobile Broadband (Satellite, early GPRS/3G vs ubiquitous 3G/EDGE and wifi)
• Compute Cluster (SUN, IBM, HPUX, etc vs Google Apps, Dreamhost, etc)
• RAID/NAS (NetApp, iSilon, etc vs Drobo, lots of other manufacturers too)
• Portable Computing (Blackberry vs iPhone, Android, Pre, Netbooks)

What's nice and interesting to note is that the Usability Fu really really matters in this zone. Corporates can afford to just suck it up and pay an expert to integrate a solution (and are almost invariably doing something weird and custom enough that they would have to even with "off the shelf" solutions). Small Business and High End Consumers don't have the time or the money to spend on Integration Experts and Solution Architects. It just has to Plug In And Work. If it doesn't work just like that, you can't sell it effectively in this price point.

Ordinary people are starting to expect computing technology to Just Work and be Easy To Use. And so they should. So, if you're in the industry at all, "It's a tricky computer thing" is not an excuse any more. It should never have been an excuse in the first place. If it's hard to use, find another supplier with a more usable product. They're starting to exist.

Comments

[thorfinn]

Two normal drobos plugged into a DroboShare will act as "one" drobo too. And a DroboShare is basically a custom Linux appliance box, so that's nice.

As for Advanced button - actually plugging the Drobo into the back of a real Linux box, you can get exactly that. There's a whole bunch of tools that let you do crap to it if you really want to.


Poor value in backups is because they were bloody hard to actually do properly - up until Time Machine. 60% of users knew they needed to do backups (bad enough), but 4% actually did regular backups (argh). Now? Anyone with a mac and an external disk or a TimeCapsule has hourly snapshot backups, and most places selling macs will push an external disk onto you for that.

Home penetration for this kind of stuff will come when Apple starts shipping machines pre-built with RAID, and Time Capsules with RAID built in, which will force everyone else to play catchup again. Probably after ZFS support makes it, which will be after ZFS is actually sufficiently bulletproof for home style random breakage, which is not yet. Essentially Home penetration will come when the tech is not just Usable but Invisible. Home users don't even want one button or any options - it just has to come along with the whole package in one unit.

Drinkies? Heh, I have zero free time - dancing 3 nights a week... OTOH, I'm in Richmond on friday nights every fortnight for dancing with an early evening before 20:30 free.

[ltempt]

While the core Drobo product might be cool, the DroboShare isn't exactly causing much excitement. Reviews are mixed, and reviews which include performance aren't good - one cited time to write a 780MByte file as "Wired gigabit Ethernet - 2:10 (6MBps)". That's not good for a device with four spindles. Actually, it would be very good for a device with a single spindle. Yes, I'm missing the point of consumer hardware. But it does mean that a 2TB Drobo would take a couple of days to write to. I would hope that the iSCSI enabled DroboPro would do wire-speed for a single workload - big performance gap.

Plugging it into a Linux box defeats the point of an appliance and turns it into a hell-like mess of high power consumption, additional management, performance bottlenecks and generally makes it more appealing to just stuff more disks in the machine. Besides, Linux .. smelly.

I doubt ZFS is going anywhere with Apple, really - they've pulled all references to ZFS from their marketing material and have been very, very quiet on the subject. ZFS isn't really cut out for consumer hardware, from a desing point - it expects I/O subsystems to honour cache barriers, for example, and that's just not a given in a land of USB storage. ZFS is nifty stuff, but it wasn't designed to be a consumer filesystem and probably never will be.

Apple's marketshare has managed to climb to around 10% - a big step up. That still leaves 90% of desktop users not able to enjoy Time Machine and such features. I'm not using with with my Mac - but I avoid keeping anything valuable on an unmirrored machine and have an rsync cronned to cover the stuff that is there. Not exactly user-friendly, though.

[thorfinn]

Yeah, DroboShare isn't the most uber of things. As you say, Linux is smelly, and the DroboShare is just a Linux box, really. :-)

As for ZFS, Apple will do something that lets them do consumer RAID in machines at the OS software level. Whether it turns out to actually be ZFS with another layer over the top, or they decide to spend a chunk of their core engineering team on designing and writing HFS++ from scratch, or they find something else, I don't know, and they probably don't either yet. My bet is they're making decisions about that as we speak.

There's some stuff in Vista Business about automated backups - does anyone know if Windows 7 has stuff in that space? I think they'd want to...