Heads in the cloud

My biggest project this year has been to scan, once and for all, hundreds of rolls of film negatives spanning the past 15 years of my life. More than 6500 pictures and 150GB of TIF files later, I am tantalisingly near the end (which will come when I figure out how to tag them all, but that's another story).

In the more immediate term, however, I have been considering how I might go about backing up such a massive amount of data, next to which my more conventional array of everyday files pales into insignificance. My short-term solution has been to copy the entire directory to an external drive, theory being that storing documents on two or three mechanical and therefore inherently unreliable devices is better than storing them on one.

With every vendor and his dog all talking about cloud computing these days, however, I have pondered more than once how I might benefit from this, which is apparently the next wave in computing. After all, the idea of getting a regular backup that's stored offsite, somewhere where it is looked after by people whose livelihoods depend on keeping my data alive, has a certain appeal to it.

But then I consider the results of projects like this — and my next one, which will involve transferring over 1TB of data from 100 videotapes before they crumble to the sands of time — and wonder whether cloud computing can really work as envisioned by Microsoft (which recently launched its Azure strategy), Google (which is all about the cloud), Amazon (whose S3 storage service provides virtually limitless storage) and nearly every other vendor, if the steady flow of press releases into my inbox is to be believed.

Consumer-focused backup services like Mozy, which automatically stores your documents in a cloud that happens to run on parent company EMC's massive storage arrays, are established and running well. But when I chatted with EMC product marketing manager Shane Moore recently about the company's Atmos cloud offering, I had another objective: to figure out how my nightmarishly large collection of photos, which I suspect isn't without precedent in the greater world out there, might be reconciled with the reality of cloud computing — and the woe that is Australian broadband.

With just 10GB of peak bandwidth per month, a full, real-time backup to the cloud would take up 15 months' worth of bandwidth at an equivalent cost of around $750 — and that's just hypothetically speaking, since speed limiting after the 10GB limit would stretch the timeframe out to years.

Obviously, this won't work. So I asked Moore how companies like EMC can throw their hats into the cloud computing ring when the model's viability depends on basically ignoring the fiscal realities of today's broadband (or, perhaps, on just taking fewer pictures than I do).

He conceded that the amount of data was not insignificant, pointing out that Mozy lets subscribers mail in a USB drive (a $150 expenditure in my case by the time shipping is factored in), where they'll copy the data and set up an account. Subsequent incremental backups will be dealt with by the Mozy software, keeping me up to date all the time.

Or will it? Sneakernet may avoid the initial, impossible economics I described above, but currently spruiked visions of cloud storage — the first practical, real-world application for cloud computing — depend heavily on a system's being always connected, and would seem to involve the transfer of an absolute flood of data whenever a large number of changes are made.

"Most people don't change their data that frequently", you will probably say, and you're probably right. But if my cloud computing backup solution works like I understand Apple's Time Machine backup works — copying changed files at regular intervals — I can only shudder to think of the bandwidth charges as my 10MB TIF files are transferred into the cloud each time I rename one of them (a common occurrence now that I'm trying to organise and tag my scanned images).

The disconnect between Australian telcos' rigid usage policies and the vision of the world's IT leaders is real and significant

If it's a bit smarter than that and only transfers the changed bits, well, that might work a bit better. But you see my point.

Cloud computing is an unprecedented opportunity for IT companies to build out a whole new charging model for storage, applications, and services. For most of us, it will create real value in interesting ways: Atmos, for example, builds on EMC's long pedigree in information lifecycle management to allow cloud computing customers to store — and index — their information according to rules they should be setting internally but usually don't.

Other vendors are looking to "the cloud" to deliver individual applications, virtual servers, entire Windows and Linux desktops that are accessible wherever an employee goes. The possibilities are endless, and they all depend on a good telecommunications infrastructure with far more generous data charging structures than Australia currently enjoys.

Could the spread of the cloud force Australian ISPs to step away from usage-based models and finally offer real, unlimited broadband packages with no hard limits? Not very likely — especially since growing use of cloud computing services will significantly boost traffic over intercontinental fibre-optic links.

Given this reality, ISPs may do well to start positioning themselves in the cloud computing space — building mirrors that let them replicate cloud data on their own local networks, then offer access to that data for free in the same way that many ISPs currently offer uncharged access to on-network video content.

Whether or not these changes will be in place in time to help with my image glut, I can't say, although I would tend towards the 'not bloody likely' end of the scale.

But, as with all things, the disconnect between Australian telecommunications' rigid usage policies and the vision of the world's IT leaders is real and significant — and, as Steve Ballmer intimated during his recent visit, plotting a path to the clouds may require some very real compromises all around.

Is cloud computing genuinely useful and exciting, or is it just a cash grab by big IT players? Do you see yourself using it?