I dont think I understand "The Cloud"
 

Is it just online file storage? If so why is everyone making such a huge deal out of it? As a geek, Im a little ashamed that this confuses me so.

google has faster servers that don't bogg down your shitty one.

http://www.spideysenses.com/wp-content/Picture%2071.png

I don't understand it either.

So its fast servers? Is it avalible to the public? Is there a subscription fee? Is that where they store my Gmail spam?

Its now a huge buzz word their using to sell everything they can..

But at one time cloud computing ment the application you were running was hosted on an online enviroment. So there was no need to install software or maintain it on your own pc.

ever use gmail? that's the cloud. ever use Google Docs? thats the cloud. Ever use Google Calendar? thats the cloud.

You store information on their servers instead of locally.

http://www.drbackup.com.au/images/Cloud_Diagram.gif?390

do a google image search for "network diagram"

the "cloud" is anything offsite that you can't possibly draw because it's too complicated.

basically equates to "the internet" but may be a private network that is just offsite as well.

now is has become a name I think for various online storage systems.

MT uses it.

There was a couple benefits for us to do so. 1 was data transfer costs. Using a CDN was cheaper than adding another server. And a lot of the load itself is caused by image requests from the users from smilies, backgrounds and even the new thread images.

Now that we use the CDN we generate about 1.5 tb of traffic each month from the actual servers, not the cloud.

2, I really cant remember what I was going to say for #2 as I started looking in the management for how much traffic in images alone we do, no dice. Forgot what I was going to say and got sidetracked.

Gotcha.

Long story short, a maximal implementation of the "cloud" concept means that both your data and the applications that you run don't reside on your own computer, they reside "somewhere else", and you don't need to worry about where that is exactly. In a particularly well-designed system, the application and data will be distributed and mirrored across many servers over a large area, and you won't know or care which particular copy you happen to get directed to every time you log on. It'll just be transparent.

Conceptually, it's not a bad idea. It means you can access your data and applications from anywhere that you have an internet connection, and from pretty much any machine regardless of whether it has a fast processor and a large storage system or not. So in theory, editing a spreadsheet or typing a word document should be just as easy from a $99 tablet as from a $500 desktop.

Gmail / Hotmail are a good example. Google apps is another. If anybody here uses Peoplesoft ERP or any similar app at work (where you access it via a web-browser), that's another.

In practice, there are some shortcomings. The networks themselves have latency, and they're not infinitely fast. So the applications are a bit laggy. Remember what it was like using BBSes at 300 / 1200 baud? Imagine typing complex documents with graphics or using large, complex spreadsheets with that kind of connection. So in reality, working from your $500 desktop will be just as slow and annoying as working from your $99 tablet.

You also can't access your data or applications from places where you don't have reliable internet access, such as most aircraft, some airports, many (most?) commuter rail systems, construction sites (this one is a common problem for me), places where an earthquake and tsunami have taken down the land-lines, etc. It also means that when the office internet connection goes down (something that happens to us once every 2-3 months), all work will grind to a halt. Fortunately, we don't rely on the cloud for our heavy stuff (AutoCAD, Solidworks, Viewlogic, etc) so we can keep doing engineery-things even while the connection is down.


Basically, there's a reason why the PC revolution happened in the 1980s. People were sick of client/server computing, where they had to log into a mainframe to do everything. Cloud proponents basically want us to move back towards that model, except that instead of the server being located in the basement of the building you are working in and connected to your desktop via a reliable, high-speed, low-latency direct connection, it will be located "somewhere else" and connected to your desktop via a slow, laggy, unreliable TCP/IP session.

What about security?

I work for the number one cloud vendor. Regardless of whether or not a cloud is public or private, there are five defining conditions for "cloud computing". These requirements are defined by NIST and are generally agreed upon by industry leaders:

1. Shared infrastructure and resources- mt.net would for instance, reside on a shared server, using a shared WAN connection with other websites.

2. Pay as you go. This means Rick likely pays only for the bandwidth he uses. OR cpu cycles, GB consumed etc.

3. User provisioned- if Rick needs more storage he can click a button a few times and get more diskspace or more servers without having to rely on other people. He can basically administer it himself without assistance from other humans.

4. Rapid scalability- if for instance, mt.net lands on the front page of CNN and there is a huge traffic spike, servers or bandwith can be quickly provisioned or decomissioned based on load. Elastic decomissioning is important so Rick doesn't have to keep paying millions for the site after it falls off CNNs homepage

5. Globally accessable- generally this means "over the internet" but it could also refer to an intranet or cipr/niprnet etc.

I can watch when y8s edits my spreadsheets. I'm living in the cloud.

What about it?

Security consists mainly of password-based authentication, similar to what you use now to access gmail or to log into this forum.

In some corporate applications, an additional token-based authentication is used. For example, I carry around a little dongle on my keychain that I have to use to access Harris resources from off-site. It has one button and a little LCD display. When I push the button, it generates a one-time-use keycode that is valid for about 60 seconds. I have to type that code into a box below my username and password in order to access my work email via the web, log into the Harris VPN, etc.

But essentially, security consists of what we're used to.

Trust is a separate matter. Do you trust that some random server (or group of servers) located "somewhere" and managed by "someone" is more secure against intrusion and theft of data than a physical hard drive in your possession?

On the one hand, laptops do get stolen, and I see no reason why someone couldn't break into my house and steal my desktop computer, or even why the cleaning staff couldn't dig through the pile of boxes and obsolete hardware in the back of my office and discover the portable hard drive that I keep stashed behind it which contains backups of all the data from my home PC and laptop.

On the other hand, I get one or two emails a year from some (bank / insurance provider / school I went to 15 years ago / DirecTV whose service I cancelled in 2004 / ...) telling me "someone (broke into our system / left a disk in the back of a cab / sabotaged us from within) and stole a bunch of data which may have included your name, address, phone number, SSN, blood type, favorite pizza topping, etc."

its like when you fart in my car on the way back from the gym. living in the cloud.

I fart getting into your car.

There are also three types of cloud services:

Platform as a Server (PAAS) - this is a coding platform. Basically you write applications using a cloud-oriented and situated programming framework

Application as a Service (AAAS) - Orcale, Siebel, Exchange Apache would be examples

Infrastructure as a Service (IAAS) - Think of it as virtual cloud hardware. Stuff like servers, storage.

Also, storing and finding vast amounts of data can be done in a cloud using flat storage/indexes and map/reduce jobs. Opposed to a database with keys, lookup tables, joins and all that. While grossly oversimplifying, relational databases were a step in the add-complexity to deal with large amounts of data and retrieve it quickly, where as cloud computing is a brute force overcome-size-problems-with-power way of doing it.

Type a word in to Amazon's search field and a map reduce job kicks off and brings back 1000's out of millions (billions?)of products from different vendors, categories, ect within a second. That's pretty impressive

Well you're talking about a specific cloud application- Amazon Simple DB and MapReduce. The cloud can be the cloud without MapReduce. That said, the functionality of MapReduce tech is pretty damn awesome.

Search engines (and the underlying database technology behind them) are cool.

Having to use a web-based application to do all of you daily administrative-type work (Peoplesoft, Siebel, etc) makes you fondly remember the days when you were using a Commodore 64 and a 300 baud acoustic modem. Except that the Commodore was faster.