Slashdot Mirror
Rackspace vs. Amazon — the Cloud Wars
fdicostanzo writes "The folks at Mixpanel are leaving Rackspace's server cloud for Amazon and have left a little note about their reasons. There's been some talk that Rackspace's offering has not been up to snuff once you scale. Analysis suggests that Rackspace's offering still has some advantages however."
It would be great if there were some cloud provider (or other name if you like), actually providing the services for VMs that we've had for years for physical machines on our desk :
-> screen access without propietary .exe (and not-windows-only) (why not via pure javascript ?)
-> a way to unplug-replug-reconfigure network
I mean even things like terramark don't provide this very basic service. Amazon is cute, if you don't mind rewriting everything in your business from scratch. And google, well Google App Engine makes amazon look flexible : proprietary API (yes - with a compatibility layer), only approved scripting languages ...
They missed the fact that RackSpace offers hybrid cloud options that Amazon just can't match at this point. Got IO issues? So did GitHub when they were running on Amazon's infrastructure. Know how they solved it? They moved to Rackspace and married the cloud for front-end with physical hardware for their IO intense workloads. It seems to me these guys may just be naive. They've probably only sidestepped their problems for now.
Even if I knew that tomorrow the world would go to pieces, I would still plant my apple tree. -Martin Luther
Interesting discussion. Perhaps more companies could make business out of their spare resources as Amazon does.
Also funny, in the comments section with GoGrid.com trolling with a $100 coupon code. Way to sweeten the pot...
Yes Google App Engine is a good example of true cloud computing. Say what you want about the limitations, but it is cloud computing.
Saying Google App Engine makes Amazon look flexible is nonsense. Apples and oranges. The upside of App Engine is that it really is cloud computing, so you actually get all of the benefits that we keep hearing about.
How could Google have done it differently and pleased you? How could they not have approved languages, when they needed to modify them to make them cloud capable? Do you expect to just toss a mysql instance and a few perl scripts into a magical cloud machine and watch it take off?
What's missing here is mention of Rackspace's recent effort with NASA on OpenStack. In short, Rackspace recently Open Sourced their Cloud Storage infrastructure, called Swift.
The problem with these solutions is they sell you services like a prepaid phone company to abstract the real cost.
My company has done the math and unless you only need the capacity say, 3 hours out of the day, EC2 (and Rackspace) simply can't compete with running your own hardware. We've heard the arguments about hiring engineers, buying servers, and renting space, but even after those expenses you still come out ahead if you have roughly more than 20 machines.
Also, Rackspace and Amazon sell Xen virtualization hosting. The software is open source and freely available if you want to use it for yourself. I just guess "Cloud Hosting" sounds better but it's not that hard to roll a similar setup if you want the scalability.
I locked myself out of a server the other day hosted on rackspace. I was able to console it from the management interface and fix the issue, not sure if I could have recovered that on ec2.
Got Code?
If you are going to post a "missive to the world" slamming someone's product, you ought to at least proofread it. It's just a bit embarrassing that the very first sentence doesn't make any sense. “...our hardware is” - yes, it's very existential hardware.
Serving as your local grammar nazi today...
Enjoy life! This is not a dress rehearsal.
Cloud Computing is such a loosely-defined and heavily abused term that its "true meaning" is almost as open to interpretation as "Web 2.0," and virtualised resources are often included in the definition.
The ever-colloquial Wikipedia states that it "typically involves over-the-Internet provision of dynamically scalable and often virtualized resources" while Foldoc states that it is "A loosely defined term for any system providing access via the Internet to processing power, storage, software or other computing services."
I'm fine with people debating the issue of the term's definition and provenance, even with people saying that one meaning is correct and another isn't, but flatly denying the existence of controversy without bothering to cite your authority is not conducive to anyone's understanding. Please, explain your position rather than simply stating it.
Meta will eat itself
The notion that there is an entity called "true cloud computing" is largely nonsense. "Cloud Computing" is just a buzzword, which covers a continuum of things: At one extreme you have essentially classical virtual private server offerings; but with an API instead of a sales rep for provisioning. At the other extreme you have fully abstracted services like "email" with absolute opacity about what goes on under the hood.
Trying to define "cloud computing" is rather like trying to come up with a good definition of "species". We know a great deal about the actual objects we are talking about; but the definition is a matter of convenience, rather than some platonic essence to be discovered.
Well, that's great and all, but if that is cloud computing, then why am I supposed to be interested ?
I want a "server in the cloud", that works preferably just like a server on my desk, except with a faster internet connection and better cooling etc.
But it still needs to do things like letting me see it's screen, replug, reconfigure the network, add/remove disks, ...
The advantages of the cloud would be : ...
-> ridiculous disk sizes possible (and for-rent - no capital cost)
-> no capital investment
-> someone else does hardware repair (and does it promptly)
-> fast scaling, that means fast access to more & bigger memory, cpu, disk,
How would that not provide the cloud benefits to me ? Perhaps you see the cloud as providing different benefits, but frankly I see little to no advantage to "automatic" scaling (and I have enough experience in the software world to understand that "automatic scaling" is about as accurate as "yes we can")
Seriously, get some good writers BESIDES yourself and get actors who can actually ACT! And while you're at it, less CGI would be good and a couple more space battles....eh...what?
Oh, 'The Cloud Wars' isn't the title of the next Star Wars movie? Oh, sorry.
He who knows best knows how little he knows. - Thomas Jefferson
It is a period of civil war. Rebel Linux admins, striking from a hidden base, have won their first victory against the evil Microsoft Empire. During the battle, GNU spies managed to steal secret plans to the Empire’s ultimate weapon, ISS, a system that brings any self-respecting admin to tears. Pursued by the Empire’s sinister agents, Tove Torvalds races home aboard her starship, custodian of the stolen plans that can save her people and restore freedom to the network.
I am officially gone from
The difference between a great idea and a billion dollars is that last 1% of the implementation - the distance between "cool" and "perfect".
I think Amazon may have bridged that gap.
I don't think that the situation is nearly as clear cut as your example.
Something like Gmail is clearly a "cloud" service. It offers email through either web, pop3, or IMAP interfaces with essentially complete abstraction of what goes on underneath.
Amazon, for its part, offers VMs, of various sizes, with essentially complete abstraction of what is underneath, accessible by means of one or more nearly frictionless programmatic ordering methods(as opposed to calling your account rep or filling out a PO or something). It is certainly a very different level of granularity, and there are definitions of "cloud computing" that could meaningfully exclude it; but there are other, equally meaningful and potentially useful, definitions that could include it.
As far as I can tell, what you're looking for is just virtual hosting, with a few specific requirements. I would think you could find all of those requirements, although I'll concede that a lot of options available today kind of suck.
"Cloud Computing" in my understanding is in fact all about the automatic scaling. I want to do a proof of concept online, and then show it to a few potential clients. I want it to basically be turned off when I'm not using it, and scale up quickly if my clients start hammering it. I only want to pay for what gets used. If I'm not using it, sell that capacity to somebody else, and keep my costs down.
If your're a full-fledged business, and you want responsiveness, and you want to guarantee a certain level of service to your clients, then cloud hosting may not be the best bet, or even the cheapest. You could still use it for special cases though. I heard a neat example: this guy needed to convert several million images into thumbnails. He wrote a little service to do it, hosted it in the cloud, let it scale way up, and churned through all the images in a few days, and it cost him a few hundred bucks on his corporate credit card. The time and expense to set up dedicated servers for this one-off task would have been ridiculous.
Like every damn thing in computer science (and really, life) cloud computing is not the solution to all of our problems, and it's also not a complete waste of time. It's a useful tool, to be used when appropriate.
Ooo! Car analogy! If you commute every day by car, you should probably buy a car. If you take the train to work and just need the car for the occasional trip to Target, then a car-sharing service might work for you.
Do you mean IIS? Or are you talking about that moon orbiting overhead?
That's no moon... It's a Space Station!
Well then "cloud computing" is simply a return to the mainframe world of old and nothing more.
A few new API's, a bigger network, somewhat better tools (thank God ! All the devils in hell couldn't make tools less useful than the old unisys mainframes - too cruel)
And of course it retains all the old downsides to mainframe computing - inflexible in the extreme (like GAE) - everything is payable, metered - huge infrastructure costs for everyone involved - totally dependant on access to centralized infrastructure - and your data is basically unsecured outside of your property - and there's no telling when, how or why the central system will go down (remember standing in line at city hall for 6 hours because "the server is down" ?)
What has changed ?
What we really need is William Shatner pitching "name your own price" for cloud resources. Yes, you CAN get cycles in a four-star data center located near the major metro area you've selected. I like the resources being commodities - more options, cheaper options. Yum!
You gotta be kidding me ... this story has been up for over an hour, and nobody has said ... "begun, the cloud war has".
I assumed that was the whole purpose of the title? :-P Or is everyone else disavowing knowledge of Episodes 2&3. ;-)
Lost at C:>. Found at C.
In the case of GAE "the server" is thousands upon thousands of small servers, not a giant mainframe that rests in a single physical location. Also, GAE isn't THAT inflexible except in the persistence portion. It's not the end all be all solution for everything. I would go for a more "conventional" solution for my random small business home page/online store. If I wanted to run weather simulations, perhaps even serious number crunching on my Facebook data, etc. etc. I might rent space out of the cloud instead of provisioning physical servers in my network. If you're doing anything at the scale at which the cloud computing stuff is really beneficial you're doing a LOT of custom code anyways so the uniqueness of GAE isn't a big deal.
Check out my lame java blog at www.javachopshop.com
It's a desirable sort of ignorance though - the whole point is a layered design so above a certain level the implementer doesn't need to think about the hardware. Sure, behind the magic curtain will be a team of people create the "cloud" abstraction, to whom it's all very concrete and not cloudy at all. Call it "ignorance" if you like, but in computer science we use the euphemism "separation of concerns" :)
In the case of GAE "the server" is thousands upon thousands of small servers, not a giant mainframe that rests in a single
By the middle of the mainframe era the same could be said about mainframes. You even had worldwide-distributed mainframes.
One thing though : their database features blew what GAE provides out of the water in terms of consistency guarantees (even if not in speed. Then again, that was 20 years ago).
If I wanted to run weather simulations, perhaps even serious number crunching on my Facebook data, etc. etc. I might rent space out of the cloud instead of provisioning physical servers in my network. If you're doing anything at the scale
I seriously doubt things like Amazon or GAE will beat even small compute clusters available at universities, both in max. capacity and (especially) price, you know, the data crunchers specially designed for the task. Especially when considering price, amazon really charges boatloads of cash for relatively trivial resources.
that crap exists. It's not cloud. You're looking for Slicehost. They have a webconsole you can use to log in, disable networking, re-enable networking, the whole bit.
I don't get why people think that "cloud" merely means "hosted VM" when, in fact, it means nothing of the sort. If system 148 of a cluster of 250 stops responding, your cloud shouldn't notice or care; it should die automatically, and get replaced automatically. If you're beholden to the health of a particular system, then it's not Cloud Computing. If you can't control and access the resources with simple REST, then it's not cloud. If it's a glass of water, it's not a cloud.
Well, that's great and all, but if that is cloud computing, then why am I supposed to be interested ?
if you're not interested, then fine - you don't need it then.
I'm not interested in buying a purse, but that doesn't mean I want purses redefined until they're something I'm interested in. I'm not interested in buying lipstick, but that doesn't mean I want to redefine lipstick until it suits my needs, or scoff at anyone else who would want to use lipstick.
That you don't have a need for it, doesn't matter. You were clearly not the target audience.
But it still needs to do things like letting me see it's screen, replug, reconfigure the network, add/remove disks, ...
No. It doesn't. That's not cloud computing. Buy a desktop, attach a monitor to it. Now you can see the screen, reconfigure the network, etc. You're thinking of a particular singular instance, and missing the *entire* point.
Sorry about replying to my own post, but unless I'm mistaken the main reason GNU was started was to get away from these big (networked) mainframe installations and the hyper-clunky and hyper-propriatary nature of their available tools. Certainly today's cloud systems have the same problem.
Apparently they had the same downsides as today's cloud services. Data goes in fine - but it doesn't come out again.
I'll start a competing service where a gnome statue touts rack space where the deer and the antelope play
I seriously doubt things like Amazon or GAE will beat even small compute clusters available at universities
That was true until recently - Amazon now has "HPC" capabilities. While not extremely impressive (yet?), you can indeed rapidly beat out any small compute cluster at a university now - at a fraction of the cost, too. Now instead of having a cluster that takes weeks to give you results, takes lots of man-hours to build and maintain, and spends only part of it's time being used...you can spin up (in minutes) a cluster that can do the work, then you can release the nodes and you're no longer paying.
Biggest issue is I have with Google cloud is it's not secure using your own domain name. You have to use google's SSL address and that doesn't fly for a lot of people.
You can't just keep scaling horizontally to avoid noisy neighbors. The problem is, unlike with cpu and memory, Amazon doesn't currently have a way to control how many IOPs one tenant has. You might even scale up from 2 "servers" to 4, and end up with the same neighbor because you're on the same underlying hardware. Plus, the issue is: it's not predictable. You might have great IOPs at one point, and then some other tenant starts consuming a bunch of them and there's contention, and your performance degrades.
"Virtual machines are not true cloud computing."
True.
But virtual machines plus on-demand self-servicing, plus elasticity on the number of deployed machines, plus per usage billing *is* cloud computing (of the IaaS kind, to be precise).
And that's exactly what they are asking for.
My startup is launching a service to do this for browsers pretty soon and the tech is flexible enough to repurpose for more typical VM tasks later. Here's a screenshot: http://i.imgur.com/UXBdJ.png The final version runs the browsers in kiosk mode to save screen real estate taken up by the window decoration and start menu. http://browserling.com/
VMs are a building block of a cloud computing environment. Definitions vary, but you can read Gartner's definition of Cloud, for example.
- Service interfaces
- Scalable and elastic
- Shared
- Metered by use
- Delivered over the Internet
So a VM is necessary but not sufficient; a VM is what you get when you virtualize an underlying resource pool. If you virtualize a pool of hardware, you get an elastic pool of shared compute resources; but there still needs to be more coordination to supply an API, metering, etc.
Beyond IaaS (Infrastructure as a Service), there are additional cloud tiers - Platform as a Service (Google Apps for example), and Software as a Service (salesforce.com, or Google Apps for your Domain).
"Especially when considering price, amazon really charges boatloads of cash for relatively trivial resources."
How they won't while there are not real competitors? Cloud computing by current standards (that's to say, those stablished by Amazon) has big entry costs but, if you really think Amazon is really so much overcharging, hey, that's your opportunity of becoming millonaire by downpricing them!
Amazon is not running a cloud with their spare resources. I believe I heard at one point that someone (Zynga?) was running 10,000 VMs on amazon at once. And that's one customer. Amazon is trying to be THE host of the future. Which is funny, since their other business is retail.
But make no mistake - 10 years from now, Amazon could easily be known as the Cloud Infrastructure provider, who also happens to do some retail. (Or less than 10.)
"Something like Gmail is clearly a "cloud" service."
Yes. Of the SaaS kind.
"Amazon, for its part"
Is cloud computing too, of the IaaS kind.
How this gets so hard to understand?
The subject says it all. WHat you're looking for is a VM - full desktop control. Cloud computing (as it's generally used) means "providing computing horsepower for your use" - the means of delivery and usage may vary. On the other hand, VMs provide pseudo-dedicated hardware equivalents for your exclusive use - including desktop use if that's what you want.
I also recall them announcing a while back that you could get preferred pricing if you were willing to let your compute cycles occur during off-hours. Sounds like a win-win.
Check out my lame java blog at www.javachopshop.com
Like spotcloud?
At least for a small instance, a reserved instance running full time for 3 years is 50% (50.8%, to be exact) of the cost of 3 years of full usage of an on demand instance. .085*24*365 = 744.6
(.03*24*365)+(350/3) = 378.666
Is the annual cost comparison.
Rackspace is generally known for great service, so I didn't hesitate to sign up and start using their cloud service for a business idea. Unfortunately, Rackspace Cloud was essentially another company that Rackspace bought and did not fully integrate or bring up to their own standards.
The several months that I had them before I migrated, I experienced:
1) Horrible technical support and the inability to get any of the actual administrators on the phone to troubleshoot. Terrible escalation procedures. If my system goes down I shouldn't have to submit a ticket and wait, I should be able to get on the phone and immediately have someone working on the issue.
2) Dropped emails. Not just queued, but dropped. Apparently if you are sending out a lot of emails, even if they are not spam (e.g. facebook automatically sends email notifications which you can elect; we had something similar), they will drop emails. Literally thousands upon thousands of emails were just deleted. No warning or notification or anything. No grace period, no buffer, nothing. This is horrible service.
3) Terrible uptime. I have had better uptime with free web hosts which I would never conceive of putting a business on...but to have to pay thousands per month for a service which is unavailable? No thanks.
4) Constantly being attacked and flooded because their security controls are not tight. We had issues sending emails, even at a low level, because someone was overloading their SMTP service because they had a virus and were sending spam. Great...so I can't send legitimate email because you can't block the account of the guy who is spamming. Nice.
5) Extremely inflexible. If you are using the Rackspace Cloud Sites, vs. installing and managing everything yourself, the interface is extremely inflexible. Simple things like ssh are unavailable. There is no normal cron. Everything is a hack if you want to get it to work, unless you simply want to use it to serve basic html and images. No decent interface for adding mail aliases. Worst offering ever. If you are going to offer an enterprise class solution, make sure you have the basics covered that even free web hosts offer first.
And what? Did they take any constructive criticisms? No. Did they give clear timelines for updates? No. Did they apologize for a lack of service? No. Credit? Not until we threatened legal action.
These guys suck. Go to Amazon.
I use Rackspace Cloud because it's simple. There's nothing to mount, nor is there a huge learning curve with setting up a VM. It's a great way to experiment with servers on a shoestring budget.
However, things change when you're moving from a handful of manually-configured VMs to an army designed to handle lots of load. Amazon's learning curve is certainly worth considering once you need to tune towards an app's specific scalability needs.
No, I will not work for your startup
Rack Space cloud sites is true cloud hosting, You do not add servers, you just pay for CPU usage like you are buying bandwidth. It is expensive and you can not customize it, but it always works and scales to meet demand.
It is not a VM, where you can install your own OS, Web server, etc. It is a apache server, mysql clustered backend and varnish cache front end.
Great for Blog hosting, PHP applications, etc. We do about 8-10 million page views on it per month and like it for what it is,
If I had staff that needed to configure servers all day, it would be different, but for hosting large dynamic LAMP websites, it is great and reliable.
Andrew
I'm a happy Rackspace Cloud customer. I use it for a few small VMs that I treat like normal, uniquely-configured servers, but I don't have to mess with all the details of running a data center, and that makes my life easier. I looked at EC2, and it became very obvious that it was not intended to be used that way. If you want to do the whole dynamic cloud thing where your log scraper uses an API to request more CPU for this VM, more RAM for that VM, and duplicate a few more web front-end hosts, EC2 definitely covers the bases, but I just wanted a couple servers with redundant power and storage, pre-built backup/restore system, in a data center that's professionally managed by people who are not me, and I wanted to do it without spending tens or hundreds of thousands of dollars up front.
In terms of business growth, EC2-style cloud computing is great for large organizations with their own developers who can afford semi-custom solutions that offload 80% of their server infrastructure to Amazon's data centers, but that's a market that will saturate quickly. The larger opportunity is customers like me, who are trying to help a small organization grow into a large organization without investing huge amounts of time and money up front (because they don't have either to spare yet), and need servers that aren't just run from someone's desk. If Amazon invests the short term returns from EC2 into something that competes directly with Rackspace Cloud, I'm sure they'll be competitive, but right now the two offerings aren't directly comparable.
There's no failure quite as dissatisfying as a complete and total solution to the wrong problem.
Anyone know what the cons of amazon cloud is at this point???
I don't want to step on any toes, but mixpanel does not seem to have the kind of traffic or growth that would call for dramatic measures (or articles). It looks like their application must be very I/O-intensive and most, if not all commercial clouds would be bad/limiting for them (does any provider give you numbers comparable to your own 10gbe or IB infrastructure without virtualization?). Sure, they can provide some room for growth on demand, but if it doesn't fit your application because you need I/O both throughput and low latency, you might still want to look at buying your own hardware.
You'll have different problems going that way, from rack temperatures to flakey RAMs, but it's much more flexible and a lot cheaper (in our case, the cost is somewhere around 15-20% of what we'd pay for EC2 including traffic over the past 5 years, as of last year or so when I bothered doing a comparison). Plus you don't need to write dramatic articles when you find out along the way that you aren't getting what you need. And you get to play with interesting hardware, but I realize that not everyone likes that. ;-)
"I love my job, but I hate talking to people like you" (Freddie Mercury)
I was unclear. I was referring to Infrastructure as a Service Cloud. (Although newservers.com, for example, bills as "bare metal cloud", and comes very close to meeting the Gartner definition.)
And GAE and Gmail are definitely cloud services - a PaaS service and a SaaS service, respectively.
But there is a difference between a VM, and an IaaS Cloud service, even though ~all IaaS services will have a VM layer underpinning them.
Amazon's "HPC" capabilities don't measure up to even a small university compute cluster. You can get 64 cores - no more (tiny ISP's have more - out of necessity).
How much can it really scale up ?
Now instead of having a cluster that takes weeks to give you results, takes lots of man-hours to build and maintain, and spends only part of it's time being used...you can spin up (in minutes) a cluster that can do the work, then you can release the nodes and you're no longer paying.
Except a compute cluster that beats the crap out of the amazon cluster nodes (due to gpu performance) costs about $1500 per node. So every company with a need for these things can easily get a 10 or 20 node cluster going. You'd have much more bandwith to the local cluster than you'd have to the amazon cloud as a bonus.
And, again, the maximum cluster you'll spin up "in minutes" is 64 cores, no more. That's pathetic.
How much cluster instance time can you get for $1500 ? About 800 hours ... just a little over a month. And keep in mind that for most calculating jobs, the GPU will beat the crap out of even an 8-core xeon, meaning 1 local node will replace 2 or 3 amazon cluster instances in raw power.
And, given that compute-intensive jobs are generally longer-running types of things (weeks at the very least), you'd run a huge profit over amazon with your local cluster pretty fast.
(and IBM's solution is a lot better for computing. At least that solution beats a small cluster. Additionally, IBM delivers consultancy that most people will need to figure out how to write efficient parallel computing programs in the first place.
I want a "server in the cloud", that works preferably just like a server on my desk, except with a faster internet connection and better cooling etc.
But it still needs to do things like letting me see it's screen, replug, reconfigure the network, add/remove disks, ...
The advantages of the cloud would be : -> ridiculous disk sizes possible (and for-rent - no capital cost) -> no capital investment -> someone else does hardware repair (and does it promptly) -> fast scaling, that means fast access to more & bigger memory, cpu, disk, ...
Congratulations, you just described Amazon AWS.
no longer working for cnet
You could ask why not refer to Google App Engine as a Platform as a Service and leave it at that, or why not call gmail a software as a service, and leave it at that.
We're really only talking about semantics, so it's not like there's a right answer, per se. But I can say that I interact with people all the time in Fortune 500 companies, talking about cloud, and IaaS is firmly considered part of "cloud services" in the vernacular.
Amazon only provides IaaS but is clearly considered cloud. Even for large entirely private enterprises, they've begun referring to internal IaaS projects as "private cloud". (e.g., I know of a large bank that ditched 30k desktops in favor of 10k rackmount servers, delivered desktops to the vast majority of employees via KMS service, and then used the unused cpu for number crunching. They refer to their 10k servers as their private cloud.)
And then there are people using IaaS and paying for it, and they call it cloud, whether that's Zynga or Netflix or the like.
So I think IaaS-as-cloud is here to stay. Which isn't to take anything away from PaaS or SaaS usage of the term.
Except a compute cluster that beats the crap out of the amazon cluster nodes (due to gpu performance) costs about $1500 per node.
No. It most certainly does not. You're leaving out the high performance networking between the nodes, the server room, the image building machine, the salary of the sysadmin that set it up and maintains it, the cost of the building the server room is in, the cost of the electricity, etc.
And, again, the maximum cluster you'll spin up "in minutes" is 64 cores, no more. That's pathetic.
All of Amazon's services have a "pathetic" starting max; we hit our limits on 3 things all at once when we started, all on the first day, and got confused for a moment...until realized you can send a request for an increased maximum, and the request is granted fairly quickly. We had our caps increased (not for HPC, but caps none the less) in about an hour. Proper planning, you request higher caps /before/ you need it.
I believe the NIST has tried to write a sensible definition of Cloud Computing, see
http://csrc.nist.gov/groups/SNS/cloud-computing/
-- Experience is a wonderful thing. It enables you to recognize a mistake when you make it again.
You know this is another complaint people had with the old mainframe software. All "power" is in the hands of the cloud provider, and you get what they choose to give, and not a bit more. The power of lots of corporations tends to get concentrated into a single huge entity like this. The mainframe "providers" maintained such an extreme lock-in that most banks are still locked in to the system, TODAY.
Why in the name of all that is good and holy do we want to return to that ? This is why GNU brought us out of "cloud computing" (then called (networked) mainframe computing) 20 years ago (and why microsoft brought the rest of the world out of it 5 years later).
Well, those who don't remember history are doomed to repeat it.
Cloud computing must fail. Not because of lack of technical merit, but because we'll see "incidents" with google and amazon cloud computing, just like there were unisys incidents 20 years ago. And the sad fact is that Stallman, in his usual way-over-the-top manner, has a point : the cloud takes away freedom.
Cloud remains very expensive for stable user loads, because there remains no means to compare compute capacity between offers. Amazon suffers no price competition. The price of Amazon's original small instance remains unchanged after four years. 30 providers might use 23 different compute metrics. See http://cloudpricecalculator.com/ as a first pass ranking of providers by mapping all providers to ECU's
http://it.slashdot.org/story/10/11/16/1549245/Cracking-Passwords-With-Amazon-EC2-GPU-Instances ;)
Your argument was already weak, given that you can request more than the initial cap, and given that you disregarded all costs other than the cost of the blade units themselves