Lidija Davis is reporting for ReadWriteWeb from the Structure 08 conference in San Francisco
In a recent report, Gartner predicted that early adopters will forgo capital expenditures, and instead purchase 40% of IT infrastructure as a service by 2011. Alistair Croll, senior analyst at Bitcurrent, and MC for the first Structure 08 conference in San Francisco, sees things differently. According to Croll it will be a lot sooner: “Right now, almost every company has someone in their IT department using the cloud to some degree.” Croll predicts that by 2009 it will no longer be almost every company; it will be 100% of companies.
Led by Alistair Croll, panelists included: Geva Perry, GigaSpaces, Jason Hoffman, Joyent, Tony Lucas, XCalibre, Lew Moorman, Rackspace, Christphoe Bisciglia, Google and Joe Weinman, AT&T.
Here are my real-time notes from the panel, followed by a primer on cloud computing.
Grid vs. Cloud
Jason talks about how grid computing is more about running multiple batch jobs. He predicts that there will be a proprietary stack and an open business model based on the cloud. Simply put, if the cloud is not billable it’s a grid; the difference is the notion of pay-per-use.
Geva insists there is room for both models. The major goal is to build applications that are platform agnostic.
Are we selling our souls – or should we not worry about lock in?
Of the panel, five deal with general purpose clouds; only one, Christophe, deals with specialized clouds.
Jason explains that in this emerging market, standards haven’t been set. This produces lock in issues, but hopefully as time goes on, we’ll have standards. Specifications are a major issue to what an open ‘loving’ cloud should do and we need to be able to reach open protocol standard and give people assurances the cloud is a good place to be.
Because we are trying to convince people that buying hardware and software is a bad idea, we need be open source and trustworthy.
Tony says that Google’s Big Table is basically a lock in and Christophe responds by pointing out that it is still just a developer preview; the theory is that it will be a very open platform.
Alistair asks what the setup is for Google; put everything in the cloud or put more out at the edge.
According to Christophe, Google has geographically distributed clusters with a lot of their services being replicated to keep customers safer. They want to ensure that if you trust them with your e-mail, the most current may be in a datacenter close to you.
What has to change for enterprise to buy into cloud computing?
Christophe says we need to allow people to do what they do best and gives the example that originally even banks were not considered safe. People kept money under mattresses. As time went on, they realized their home could also be broken into, so storing it at a bank is a smarter solution. He expects that over the next couple of years we’ll see companies keeping data more safe, and enterprise more willing to buy in.
Joe thinks benefits will come from a commitment to the enterprise space and ensuring the infrastructure is secure.
Geva is concerned that companies don’t want quick credit card access as afforded with most cloud computing solutions. They want contracts and support; they want help. It needs to be given. Lew thinks the definition the concept is very confusing for enterprises. The word cloud is starting to mean less and less, yet at the same time it is cheap easy and it’s getting into enterprise through the back door (IT departments are beginning to investigate).
The session went well and was held before a standing room only audience.
Cloud Computing: A Primer
The ‘cloud’ is a well known metaphor for the Internet. IT folk have been drawing puffy blobs to represent it for decades. The latest buzzword ‘cloud computing’ however, is not as well defined, and not as familiar.
However, the general consensus is that cloud computing has become an umbrella term covering various solutions. Here’s a basic rundown:
SaaS
Software as a Service is a subscription based application that is hosted on a remote server, and accessed through the Internet. Subscription based because you don’t pay a licensing fee–as long as you’re subscribed (whether the service is paid or not) you have access to it. Accessed through the Internet because the software is not downloaded onto your computer; the program and the data is accessed via the Web. Think Gmail as the basic SaaS, salesforce.com as the larger model.
IaaS
Infrastructure as a Service offers computing services on the Internet designed to make web-scale computing easier and more cost effective. You no longer need to buy servers, network equipment, memory, CPU, disk space or data center facilities; they are provided as a monthly service, generally based on capacity used.
Given there is no up front capital to purchase hardware, software, or facilities, corporations and developers are quickly moving to this a smarter option. Amazon’s Elastic Compute Cloud, or EC2, is a prime example of IaaS.
PaaS
The idea behind Platform as a Service is to deliver development environments as a service; offering programmers everything they need to create applications via the Web – including building, testing, deploying and hosting services – without the need to ever install anything on their own computers.
Examples of PaaS: Google App Engine, Yahoo Pipes, and Force.com from SalesForce.com
