Post a Comment
To me the worst aspect of the "cloud" as it is sold now is that you are wasting perfectly powerful computers with plenty of storage as thin clients of overloaded servers. I guess storage makes more sense, but only for backups.
Let's not forget that the "cloud" was standard practice once, and it lost to personal computers.
Wake me up when they get more aggressive and say start selling company controlled "cheap" computers that would load a synced environment for that individual user but at the same time use "his" computer as a backup server for other people's data. They only need P2P technology, DRM, buzzwords and a shitload of cash.
Umm that sounds a lot like Google Chrome only more evil.
The only app, I've seen, that really takes advantage of the cloud is Panda's Cloud AV. It makes AV scanning a group effort. It's like FOSS development for virus scanning, and the concept is really cool from a security stand point.
http://www.cloudantivirus.com/en/
Zonbu is already kind of doing that. You can buy a thin client from them, pre-configured system, or use your own system to access a virtual desktop.
http://www.zonbu.com/whatiszonbu/
There are many legal issues, user issues, and licensing issues which prevent cloud computing from living up to your expectations. For instance, botnets could be considered a type of cloud computing, and we are all aware of how those are used.
for argument sake, lets call the opposite of "cloud computing" "ground computing". "Ground computing" then means running an OS installed on the ground machine and let it process data stored on a same machine and then store the processed data back to the same machine, everything is basically done locally.
"Cloud computing" to the most part means the OS itself or the data to be processed or processed data or the crunching of data occurs somewhere else on the internet, on someone else's hardware.
Cloud computing is a bit problematic to free software advocates. You cant really say you are a king of your own data if it sits on somebody else's hard drive somewhere on the net. How is that different from having the same data locally but in an proprietary format?
Data format and data location are different enough. The issue is trust and accesibility. Trust [towards the provider] that it stores your data losslessly (no data loss), and securely (nobody else can access your data). Trust that it doesn't go bankrupt and disappear. Trust that it provides constant accesibility, i.e. no service stops because of maintenance or power outage, no service outages whatsoever. Trust in your ISP that you have constant internet access, no interruptions whatsoever. Trust in every other ISP wheverer you may travel, that it also provides constant high rate connections towards your data provider.
That's a lot of trust.
There's a long, long way to go till companies can provide that level of trust.
I believe David's point is that there are two kinds of usage for the term "cloud computing" and the one everyone is wary of (thin clients, hosted offerings, software as a service, etc.) is quite different from the one that's already gaining lots of popularity (Using "private clouds" to make provisioning much simpler and minimize waste).
Edited 2010-09-07 22:57 UTC
Cloud computing has it's merits, but it can also be a trap, if you're not careful. Delivering your data to someone outside of your physical scope means that, sooner or later, you will have problems with your data (as always), but you will be unable to step up and deal with it yourself. You'll be at the mercy of some other company and that is ALWAYS a bad thing, because you're just another customer, your data isn't as valuable to them as it is valuable to you.
None the less, cloud computing is excellent for volatile things. For example, a marketing company can rent computing and bandwidth resources on a cloud computing vendor to deal with a summer campaign for two weeks, instead of spending money on new servers. It's a good thing when you have a scalable IT architecture.
Also, cloud computing is just a fancy-just-invented-now-so-it-must-be-good name for grid computing. I yawn when I head "cloud computing".
The problem with cloud computing is the same as when IBM introduce the concept (though they didn't call it cloud computing) in the 1950's and the solution is the same: if it's critical to your business, do it yourself regardless of cost. Otherwise, go with the most-reliable, cheapest solution.
When it comes to hosting very few businesses need more than a VPS.
Managed dedicated hosting is nothing new, you get a single instance and the hosting company handles the rest.
Cloud computing is more useful to companies that require a heavy amount of shared processing. But even in many of those cases it is not going to be cost effective given how cheap gigabit and multi-core cpus have become.
Cloud computing is supposed to bring in the age of the web app but people said the same thing about RoR, Java and AJAX.
Cloud based backup services are useful but that is nothing new either.
I saw an endorsement somewhere for cloud computing that stated small businesses now have access to the same level of data centers as fortune 500 companies! And why the hell would the typical small business need access to that type of data center?
So overall I think cloud computing is mostly hype even though it is useful in some cases. If it gets popular dedicated hosting companies still have a lot of room to cut their prices.
I just clicked on an ad here for cloud hosting at the planet.
The offer is $50 a month for which includes 1 vcpu. Well how much power is one vcpu? If it is 1/total_instances that means nothing to me. For $180 I get 2 vcpus. When will I find out if I need 2 vcpus? When the company arbitrarily tells me? Oh joy, just like shared hosting.
Much like real clouds, the white puffy kind is inherently much like cloud computing. The sad thing is there are always a group of marginally employed tech assholes that will buy into the latest fad in IT and repeat it over and over again and eventually the community moves on to the next idiot stage of its existence... "The next big thing."
...because "cloud computing" is simply distributed systems. Not a bad idea, nor particularly revolutionary. The main difference today is that the logical servers can be slightly more abstracted from the physical servers than previous.
Besides that - my data will be stored locally, thank you very much 
I for one love virtualization. I actually wrote a big thank you blog at work praising our sys admin guys. After reading this article, I think a lot of it is due to virtualization. We are fully virtualized. Every test server... everything. I want something up... boom it's there in 1/2 hour.
I've always wanted to just have my work desktop hosted as well and I just vnc/remote desktop in. They give us powerful laptops to work with... but I just use it as a thin client to my desktop.
That all said, what I still don't see... and what I don't want is... web apps. They are good for certain things. Yet, I sometimes have to fight to say... 'not everything should be a webapp'. The biggest problem is that you have to then host the application on some webserver. That's great if you as a company are willing to devote the time, money, and resources to support that. Yet, a lot of the time, you just don't have that and a local program is much more usable.
You do end up with more complex problems with webapps... especially scaling it all out.
You have to be really careful with apps that really are not collaborative or web dependent, but you think its going to be easy to deploy... so you make it a webapp. 9 times out of 10 you'll end up making things more complicated.
Granted, most of our customers are still 'techy' so that might be a different perspective... they have no problems running an exe or something.
Even for smartphones, the local client tends to win. The 'web' was supposed to be do anything anywhere, but it's really poor as a programming and integration environment.
Hi, you Worte "There are two technological innovations that, available together, make cloud computing possible:"
I totally agree but want to add that without service orientation inside of the apps in the cloud you won't have the right services. So the ROI won't be as high as if you have those services. So, you need a former hype "innovation" - SOA - which is mainstream today
There is one of my Blog items "Cloud-SOA=Zero" at
http://www.x-integrate.com if you want to read more about this topic.
Is it Skynet then?
Business use cases are being provided because we are talking about whether or not it is useful to businesses.
Furthermore the business use cases described here fit the services offered by Amazon's EC2. Perhaps Amazon doesn't understand cloud computing either?
I've examined the issue of cloud infrastructure implementation and management from both sides - from the point of view of those deploying solutions as well as those consuming their services. My final determination is:
I really don't know clouds, at all.....
Humor an old man!
Or maybe what you own. I like the cloud for some things, like my address book or my my email, and generic Apps with a nice AJAX interface, But I have several usb drives (16 - 32 GB) that sit on my keys and are usually rsync'd to a folder in ~/Key_Drive - It may be insecure and possibly innefficient or it may be perfect, because I cannot do something trusted on 'your' machine. And I do not expect trust on 'Their' Data Center. If I lose my key_drive, it is a $30 commodity loss and I get full ROI by the third or fourth usage. None of what I do publicly is expected to be private. unless I elect to encrypt it - which triggers many U.S. Government red flags (and only if I mail you my 'key' or put it on a key drive and then hand deliver it to you I can't really expect privacy or security)
I guess for the user the Cloud is New and Great and (dare I say) Sexy but it will always feel like a thin client on a leaky network.



