But they're in control now.

Because, like, they outsourced their servers to the cloud, and their developers to India, and the helpdesk too, and, you know, they all chose that entirely by themselves. Well, their managers did that, but if the users complain they too get outsourced along with the IP that makes up the secret sauce that's supposed to spin money for the company. Progress, see?

I've seen Matt drift ever more into this cloud thing, to the point that he really should be upgrading to a space suit soonish. In that, he's probably very in with the hippest of the silly valley brass. But what it's got to do with Getting Stuff Done while keeping your secrets, well, secret, he hasn't managed to make very clear to me.

If there's an upside to this, it's that there'll be good monies in selling bandwidth, and maybe caching appliances that keep up appearances while the cloud's out, where before there'd've been the same hardware running the service locally. Buzzword compliancy, it's serious business, I'm telling you.

Space suit...purchased!

I had it made in the cloud. :-)

There IS an awful lot of really bad IT out there.

Been not really watching another university trainwreck from afar, where the departmental IT got sanitized away entirely and every computer-y device got put under centralised control, classified as one of three flavours of windows desktop install. 'Sright, non-desktops don't exist outside central IT. The CS dept. didn't mind, which told me nothing I didn't already know from the days I actually walked around there. Departments like physics had a bit more trouble. What sort of windows desktop class is your old electron microscope, or your old super, or any old research-class lab equipment, really? So they quickly reinvented the "DIY" class, for the now-nonexistent local IT to sort out.

They'd also deprived the departments of their own sub-domains (that had been in existence since, well, ever) basically because they couldn't make sexchange to DTRT with it. Yes, really, that's what "technical reasons" means over there. This a well-known "technical" dot-nl university.

Anyhow, there are still Good Reasons why you'd want to have IT departments, and SaaS, cloud, whatever don't change that. Though some will have grabbed control for the heck of it, just like there are IT depts. that do every desktop install entirely from scratch and that do not understand DNS, much less are able to run an internal server, most will have done so because they got sick and tired of yet another done-it-themselves misconfigured desktop install and the associated whining that clogs up the ticket queues. Alright, maybe SaaS and cloud will manage to make the first level IT bod entirely unable to fix the problem, leaving him no more leverage than getting into shouting matches with your friendly SaaS/cloud provider's FAQ AI^Wsmall shell script, or whatever.

Personally I wouldn't mind playing ISP of sorts, though on the understanding that this means providing connectivity and no more; I won't fix your SaaS, nor your cloud, nor your computer. I'll just turn off your port should my abuse phone start ratting on you. If all I have is building granularity, then yes, the entire building will go. For if I don't, my upstream might turn the entire site off, which would be worse. Thus the need for someone who can deal with that sort of thing locally arises.

Then again, I'm not much of an empire builder. Empire building merely means the chiefs aren't thinking of the entire company, meaning they're not doing their job. That makes it a failure of management, of the brass even, and failing IT support is but a scapegoat. Should've put the right people in the right places; chiefs who know how to get their indians what they need, be it IT support, the right information to provide that IT support, or whatever else they need.

Of course we'll still need IT

It's just that the nature of it will change. That's one of the things I was trying to communicate (clearly not very well). Thanks for reading and commenting.

The changing nature of IT

Where we used to say "we have an ETA of X for Y service to be recovered, and the initial analysis of the failure points to Y as the root cause"...

We now get to say "our vendor finally confirmed there is a problem with X, they didn't tell us what their techs are seeing but we have an ETA of Y for the next update, and once the outage is over it will be Z days before we get an RCA".

The cloud does not change life's basic tenants - like shit happens, and you get what you pay for. I can, at least, say with confidence that we have made informed decisions about what we've moved, to where, and how... but I doubt most companies that have made this move have done so.

That's called "projection".

It's a rather nefarious kind of wishful thinking. You're saying amazon will learn because they will want to avoid losing custom through embarrassing gaffes, essentially because you believe they should. That is actually quite an assumption there. It's also worded wrong; they're private companies even if they're traded in public. They're not accountable to the public. "Face" works only as well as it is valued, which isn't all that much outside Japan.

Maybe amazon will learn, though they've already shown they've quite a lot to learn yet, especially for something their size which to me indicated the cloud operation started out as and largely still is a corporate hobby of sorts, but I've also seen cloud-y providers like google have repeated outages without any sort of redress whatsoever. And that's not just the free parts of gmail, either.

More importantly, if "cloud wins the day" then there will be a proliferation of fly-by-night cheaper outfits that plenty of start-ups will use, some of which will grow and perhaps find they've unwittingly handed over the crown jewels and can't get them out again. Perhaps they'll learn the consequences of that the hard way, should the provider go bust, say.

But before that your business continuity planning now must include a trustworthyness assessment of your cloud provider based on your best guess as to what they believe their public standing should be and how much effort they're willing to put in making that happen through dilligent care for the technology that hosts your data, under their control. Assuming it is under their control and not, say, outsourced again to the cheapest on-line bulk storage provider. Personally I prefer MTTF figures to guesstimate with, but maybe that's just me.

Sysadmins have learned the hard way to not trust anything that isn't backed up by something they do trust--to the point that "slightly paranoid" is what the job *needs* you to be. That something might be tape, or it might be contracts and a small army of lawyers. Though even the latter cannot bring back your data if your cloud provider forgot to use the former.

And then there's the thing that even if "market forces" will manage to make it all good (for the usual industry standard mediocre value of "good"), the process to get there will include plenty of painful gaffes, spectacular data losses, embarrasing goofs, and other lows. From a business continuity PoV, at least I would like to avoid betting the company on all that.