There's an awful lot of scepticism on slashdot about the cloud, which is healthy in a way, but I think in general people are hugely underestimating the impact that cloud computing is going to have on IT deployment. It is going to affect us all; software as a service holds huge challenges for the free software movement, some skillsets such as traditional IT support are not going to be as useful, and the way we write software is going to change further.

I'm no cloud zealot, I've just been reading about it a lot and talking to Cloud providers (some large, some medium-sized) and academic experts. I've tried to answer the many points that have been brought up here:

-- "We've been down this road before."

We have, but things *are* different now. Firstly, we have sophisticated and mature virtualisation technologies that allow efficient coresidency and management of VMs. Costs per CPU hour have dropped. Internet access is incredibly pervasive. The "post-PC" era of tablets and smart phones are producing a huge demand for cloud-based storage and services. Does this mean cloud will automatically be successful? No. Does it mean that comparisons with previous era's are not necessarily correct? Yes. If you want another example, tablets didn't 'work' in the past... but now they do.

-- Moving to the cloud won't change anything.

Yes, and no. We will still need IT to manage the cloud services, and engineer bespoke cloud products. Users will still require support. But you're no longer talking about rolling out O/S updates across your company, or installing the latest version of Word. No more capital investment in some server hardware, no more long-term planning of purchases of those servers. If a thin client is broken, you just replace it, and maintaining those thin clients is a hell of a lot easier if they're dumb.

-- Bespoke solution X won't work on the cloud.

No, it won't. But your Exchange server certainly can be moved to the cloud quite easily. In fact, many companies start their move to the cloud with Exchange, and then migrate to live apps... the point is, that you don't have to move *everything* to the cloud in order to make savings and find other benefits.

My advice is, go learn about cloud computing, start looking at the architectures that cloud applications use. Read up on Amazon Web Services and try it out. Take a look at Google App Engine. Read a few books looking at the business case for the cloud before you dismiss it.

RS

> And it's *always* cheaper to in-source (provided you can find the appropriate resources). You can either do it yourself, or you can pay someone their cost, which could be your cost, plus 20% or more overhead and profit. So outsourcing costs you a minimum of 20% more than doing it in house. But all the consultants swear it's better to outsource - to their company. That's like hiring the Fox and Co security company to guard the hen house.

Simply untrue. I'm an academic writing and reading a lot about cloud computing right now, and I can tell you that studies and anecdotal reports both show that outsourcing in its traditional and cloud-based form are both cheaper - *provided* that you are starting from scratch. Migration is another issue entirely.

RS

I really don't think you're being balanced here, although the previous poster wasn't either. No need to flip the other way (I am a Mac / Linux user):

> You can run anything you like on any Macbook. And as other responders have noted it DOES solve the "too many models" problem, there is a limited selection of base models. It's basically, do you want an air or not and how much memory (RAM and storage) with reasonable defaults so you cannot go wrong simply ordering default configurations.

Yes and no. In reality, things that work and/or look nice or work well on Linux, just don't cut it on a Mac. For example, Inkscape.

> If you help someone buy a Mac you can simply say "try the Apple store Genius Bar first". Chances are they can help, they generally are pretty competent and pretty technical.

Yeah, I do this too. But going to a store can be a pain (for example, if you live too far away from your nearest store, or it is very busy). The last city I lived in, the nearest Apple store was 70 miles away. And you're not going to visit the store for every small problem you have.

>> You still have software problems with upgrades.
> Actually that is REALLY rare, and again - Apple Store, not you. I have not had upgrade problems in years with any Mac software updates.

Really, this is not rare at all. Particularly with the last OS X upgrade, many people I know had lots of different problems, for example with Wifi and networking.

>> People still need to search some forum from time-to-time to figure out how to fix some strange new issue.
> Or have someone help them for free, at an Apple Store. Are you starting to get what makes a Mac such a great idea for non-technical users? How is a non-technical person supposed to search forums for "odd problems" anyway? They cannot.

In reality, Mac users I know, even those who aren't geeks, do spend time searching the web and looking through forums to solve their problems.

>> Ever try and build something from Ports only to have it *not* friggin work when you upgrade?
> Come on man, that's not something the original poster is going to find a problem. If you are compiling UNIX utilities yes you may have some hiccups, but even then you can usually just recompile!

Honestly, if you're doing techie hacking, the reality is you can't beat a Linux box. The main reason is that most UNIX hackers run Linux, so it's the first thing they support or test. Of course, if you're not a geek, then this doesn't matter to you.

>> Apple has gone to great lengths to make the use experienced top-notch, but it still has it's problems just like Windows
> That is the biggest misconception. I still have to help people with modern (Windows 7) pc's from time to time. Macs do NOT have problems anywhere close to what non-technical windows users have every day.

Probably not, but they still do have plenty of problems. OSX has lots of bugs and quite a few poor UI design decision. Linux and Windows also have lots of bugs, although Linux distros have a much worse UI in general.

> To even think about buying a non-technical user a Windows box and all the inherent baggage that comes with it is simply cruel to my mind - and it's a cruelty that as I said will be punishing you as well as the recipient.

I agree that Macs are generally a better choice than Window boxes, although you've got to take cost into account.

RS

> From all the gripes I see the problem is finding the little buggers.

> Has anyone here actually held one in his sweaty hand?

I registered with RS at 10am on the day of launch. They neglected to notify the mailing list that it was launching, so I was late. I finally got to order just over a week ago, and was told it should arrive within 3 weeks. I think they're close to clearing the backlog of orders, as the Pi's are now in serial production.

By my reckoning, you should be able to order sometime next month without a place in the queue.

RS

Parent modded 'Funny'; actually, was a slight exaggeration... it would be 101.

> We've accomplished all the easy, flashy stuff.

Landing on the moon was easy?!

Oracle will probably sell her column off anyway, or at least take someone to court over it.

> Java is deprecated.

What?

I'm in the UK, not the US. I'm a post-doc in CS.

As many others have pointed out, fraud is rare. It probably makes for good headlines, but I have never personally seen or heard of an instance of fraud.

The negative side of academia (for me) is the growing trend of heavy focus on metrics, e.g. the number of papers published (sometimes weighted by impact factor), or the number of citations a paper receives. This is generally a result of top-down meddling by the government and distracts scientists from doing good work. I disagree with it and try to avoid thinking about this game.

I have not chased publications, but my career is going just fine.

Some people claim that you have to work 60+ hour weeks etc. to be a successful academic. In my experience, this isn't true. Some people do work those hours, but most of them have poor time management skills and are not very productive. I don't see anyone working significantly more hours than myself achieving more over an extended period. They tend to get tired, burn out a bit, and spend a lot of time procrastinating. You can be a successful academic by working reasonable office hours, plus the extra hour here or there, maybe a weekend day occasionally, and then a little extra around deadlines. You just need to be organised and disciplined.

So, don't believe all the negativity you read. I love my career, and I love the flexibility that academia gives me. I work with tons of creative people in a relaxed environment where I set my own hours, have a large say in what I do, and get to work with intelligent students who keep me on my toes!

Oh, the money sucks ;-) It's enough to comfortably live on, but compared to industry (and given the sacrifice of doing a PhD in the first place) it is a serious financial sacrifice. I am considering going back into industry at some point in order to build some savings.

Hope that's of use.

RS

Good point - you've hit on a major concern of Cloud migration.

There are two answers to this scenario, the first being "the law needs to catch up with the tech", so that this situation could not arise, and the second being "anything that is critical, e.g. IP, should be stored off-cloud, and accessed only when necessary via secure communication". Encryption is your friend - and there are ways of ensuring that encryption keys will only be available if your company is sued, etc. (i.e. the situation we have right now).

RS

"Firstly, the schedule for manufacture for every UK business we approached was between 12 and 14 weeks (compared to a 3-4 week turnaround in the Far East)."
http://www.raspberrypi.org/archives/509
[Posted on January 10th, 8.5 weeks ago - and manufacturing had already started at that point].

I guess UK manufacturing wouldn't have been much slower, after all.

RS

> So sorry if Bitcoin is flawed to the point where it can be so easily stolen by little old root

This is a really good point. Why is such an important password even known by anyone? If it is a system-wide password that can be used in the way it has, then you might think its very existence is a bad idea. Use public/private key authentication instead, if this account *must* exist, and physically print those keys out, possibly in parts, and distribute them in extremely safe locations (security deposit boxes, for example) just in case the account ever needs to be used. It should not be required on a day-to-day basis. Even if you make the mistake of allowing it to be used in that way, then the people with access to that information should be very few and highly vetted.

RS

> Those people had no business storing $15,000 worth of irreplaceable data, electronic currency or not, on a service with these kinds of terms. Instead of spending an appropriate amount of money for the proper security they gambled with a service not designed to insure against that kind of liability and lost.

This is absolutely correct. I'm reading the most basic introductory books to cloud architectures at the moment, and this is constantly brought up. If you have vital data like this, store it encrypted and off-cloud. If you have to use it within your application servers, either fetch it from the off-cloud location, or else use encryption and distribute the data so that a single compromise (e.g. of one cloud provider) cannot result in a complete compromise of your data.

Linode should pay up, but the users are (for once) to blame.

RS

You may find use for this:

http://www.simplynoise.com/