Archive for the ‘Cloud computing’ Category

Why IT needs to get over cloud-aversion

Thursday, April 10th, 2008

The cloud was everywhere, except in the Las Vegas skies, at the Gartner ITXpo conference this week. The cloud computing sessions were full and plentiful, yet the mood was hardly energetic.

Several Gartner speakers exhorted the mainly-IT audience to accept the cloud as an unstoppable force, and not ignore it or oppose it. Audience questions showed that IT is not yet convinced. Most questions were tactical, such as how to handle integration or assure security -- worthwhile goals in some circumstances, but not the key issues in thinking of cloud computing, notes analyst Darryl Plummer.

Cloud computing, and external sourcing of all types, is inevitable

That cautious reaction may be natural but it's also dangeous, as the cloud is coming. Already, Gartner analyst Ben Pring notes, outsourcing is the No. 2 expense (30 percent of IT budgets), not much less than applications (32 percent). In 2011, Gartner predicts that outsourcing will rise to the top expense spot at 33 percent, while application spending stays at 32 percent and thus falls to the second spot.

Of course, that outsourcing spend includes everything not done by IT, including consultants. The cloud portion -- including software as-a-service, Internet-provisioned infrastructure and cloud-based app platforms -- is a tiny percentage today.

But Pring's message is that companies are moving away decisively from trying to own everything they do, instead keeping what is strategic (meaning what makes them better than the competitor) and having the rest handled generically. That will mean contracting, the use of open source, hosting, business process outsourcing and the various forms of cloud computing.

Gartner calls all of these "externalizing," with the cloud being the key way to externalize IT-delivered processes.

The trap of overengineering

Plummer made a similar argument, plus suggested that IT's insistence on complex technology approaches will blow up in its face. He said that IT efforts are typically overengineered, "built for problems that very few companies actually experience. ... The old model is pulling us down. IT is trapped in overengineering."

He points to the evolution of SOA, which started as a simple approach to decoupling services so they could be delivered flexibly and speedily. Pretty soon, complex messaging buses showed up in vendors' hands, coupled with lots and lots of new standards that have limited many of SOA's strengths. "WS-* is too complicated," he says. "Web services and REST services are all you need for most processes." Yet the industry -- and IT -- have gotten happily enmeshed in more complex approaches such as SOAP. "REST will prove to be more successful than SOAP," he says.

Plummer contrasts that to what is happening in the cloud. Providers such as Salesforce.com, Google and Amazon.com use simpler protocols. For example, a typical IT approach to transaction processing would require a rollback mechanism if a transaction doesn't complete, a complex approach that doesn't scale easily, he says. So what do PayPal, Amazon.com and others do? Avoid rollback and simply use a compensation approach (a credit or refund) that is easier to implement and scales much more easily.

Likewise, simple HTTP works quite well as the backbone of Internet communication, while the complicated alternatives proposed in the 1990s went nowhere, Plummer says.

How to succeed in a cloud world

So what does this have to do with the cloud? In a word, simplicity. "Externalizing" nonstrategic processes, whether business or IT, is going to happen. A few jobs will remain for the really complex stuff that needs the multiple layers of security, high degree of redundancy, lightning-fast computational capability and so on. But most of IT will be about orchestrating a combination of internal and external resources.

In that world, overengineering is out and fit-to-purpose is in. In that world, complexity is out and simplicity (and genericity) is in. Value comes from orchestrating, not rebuilding. Integration without imposing unneeded dependencies is in.

In other words, IT will have to apply its combination of creative problem solving and engineering rigor in a new context, where the economic considerations are one of the key metrics. This approach brings IT into the business, notes analyst Richard Hunter, and gets rid of the silly "IT/business alignment" concept that has perpetuated IT being about something other than the business.

Cloud computing is a symptom of a larger change going on, one that may seem scary to IT traditionalists, but that holds the promise of making IT about solving meaningful problems.

Why IT needs to get over cloud-aversion

Thursday, April 10th, 2008

The cloud was everywhere, except in the Las Vegas skies, at the Gartner ITXpo conference this week. The cloud computing sessions were full and plentiful, yet the mood was hardly energetic.

Several Gartner speakers exhorted the mainly-IT audience to accept the cloud as an unstoppable force, and not ignore it or oppose it. Audience questions showed that IT is not yet convinced. Most questions were tactical, such as how to handle integration or assure security -- worthwhile goals in some circumstances, but not the key issues in thinking of cloud computing, notes analyst Darryl Plummer.

Cloud computing, and external sourcing of all types, is inevitable

That cautious reaction may be natural but it's also dangeous, as the cloud is coming. Already, Gartner analyst Ben Pring notes, outsourcing is the No. 2 expense (30 percent of IT budgets), not much less than applications (32 percent). In 2011, Gartner predicts that outsourcing will rise to the top expense spot at 33 percent, while application spending stays at 32 percent and thus falls to the second spot.

Of course, that outsourcing spend includes everything not done by IT, including consultants. The cloud portion -- including software as-a-service, Internet-provisioned infrastructure and cloud-based app platforms -- is a tiny percentage today.

But Pring's message is that companies are moving away decisively from trying to own everything they do, instead keeping what is strategic (meaning what makes them better than the competitor) and having the rest handled generically. That will mean contracting, the use of open source, hosting, business process outsourcing and the various forms of cloud computing.

Gartner calls all of these "externalizing," with the cloud being the key way to externalize IT-delivered processes.

The trap of overengineering

Plummer made a similar argument, plus suggested that IT's insistence on complex technology approaches will blow up in its face. He said that IT efforts are typically overengineered, "built for problems that very few companies actually experience. ... The old model is pulling us down. IT is trapped in overengineering."

He points to the evolution of SOA, which started as a simple approach to decoupling services so they could be delivered flexibly and speedily. Pretty soon, complex messaging buses showed up in vendors' hands, coupled with lots and lots of new standards that have limited many of SOA's strengths. "WS-* is too complicated," he says. "Web services and REST services are all you need for most processes." Yet the industry -- and IT -- have gotten happily enmeshed in more complex approaches such as SOAP. "REST will prove to be more successful than SOAP," he says.

Plummer contrasts that to what is happening in the cloud. Providers such as Salesforce.com, Google and Amazon.com use simpler protocols. For example, a typical IT approach to transaction processing would require a rollback mechanism if a transaction doesn't complete, a complex approach that doesn't scale easily, he says. So what do PayPal, Amazon.com and others do? Avoid rollback and simply use a compensation approach (a credit or refund) that is easier to implement and scales much more easily.

Likewise, simple HTTP works quite well as the backbone of Internet communication, while the complicated alternatives proposed in the 1990s went nowhere, Plummer says.

How to succeed in a cloud world

So what does this have to do with the cloud? In a word, simplicity. "Externalizing" nonstrategic processes, whether business or IT, is going to happen. A few jobs will remain for the really complex stuff that needs the multiple layers of security, high degree of redundancy, lightning-fast computational capability and so on. But most of IT will be about orchestrating a combination of internal and external resources.

In that world, overengineering is out and fit-to-purpose is in. In that world, complexity is out and simplicity (and genericity) is in. Value comes from orchestrating, not rebuilding. Integration without imposing unneeded dependencies is in.

In other words, IT will have to apply its combination of creative problem solving and engineering rigor in a new context, where the economic considerations are one of the key metrics. This approach brings IT into the business, notes analyst Richard Hunter, and gets rid of the silly "IT/business alignment" concept that has perpetuated IT being about something other than the business.

Cloud computing is a symptom of a larger change going on, one that may seem scary to IT traditionalists, but that holds the promise of making IT about solving meaningful problems.

Google launches app-hosting service

Tuesday, April 8th, 2008

Google tonight stepped into the burgeoning cloud computing game, launching Google App Engine, a Web services initiative that allows developers to host applications on Google's infrastructure.

Available free to the first 10,000 developers, the service taps a fully scalable Web platform with Python-based application servers, GFS (Google File System), and BigTable data storage system at its core.

It is a move that competes obliquely with Amazon's well-established AWS (Amazon Web Services) platform, with Google offering AppEngine as a tightly bundled package and Amazon serving its S3 storage, EC2 (Elastic Compute Cloud), and Simple DB database a la carte.

Considered by many to be a necessary entrant for legitimizing the nascent cloud computing model, Google will allow developers the opportunity to serve up to three applications on App Engine, each with a cap of 500MB of storage and 10GB of bandwidth per day, which will serve an estimated 5 million pages per month.

In other words, don't expect full-blown enterprise-class utility computing from Google off the bat. Instead, single-developer widget shops and bootstrap startups will be among those to first test Google's initial foray into the infrastructure-for-hire space.

[ For an in-depth look at cloud computing, see What cloud computing really means ]

Google launches app-hosting service

Tuesday, April 8th, 2008

Google tonight stepped into the burgeoning cloud computing game, launching Google App Engine, a Web services initiative that allows developers to host applications on Google's infrastructure.

Available free to the first 10,000 developers, the service taps a fully scalable Web platform with Python-based application servers, GFS (Google File System), and BigTable data storage system at its core.

It is a move that competes obliquely with Amazon's well-established AWS (Amazon Web Services) platform, with Google offering AppEngine as a tightly bundled package and Amazon serving its S3 storage, EC2 (Elastic Compute Cloud), and Simple DB database a la carte.

Considered by many to be a necessary entrant for legitimizing the nascent cloud computing model, Google will allow developers the opportunity to serve up to three applications on App Engine, each with a cap of 500MB of storage and 10GB of bandwidth per day, which will serve an estimated 5 million pages per month.

In other words, don't expect full-blown enterprise-class utility computing from Google off the bat. Instead, single-developer widget shops and bootstrap startups will be among those to first test Google's initial foray into the infrastructure-for-hire space.

[ For an in-depth look at cloud computing, see What cloud computing really means ]


eXTReMe Tracker