Adventures In Utility Computing

The business of utility computing.

Monday, April 09, 2007

CRB ruling, an impediment to internet radio adoption

It has been a while since I have posted. While utility computing technology continues to march along I have turned my personal focus to the structural impediments to broad utility computing adoption. The best example is the recent Copyright Review Board ruling on web-cast royalty rates.

This ruling, applied retroactively, increases streaming royalties by over 130% over the next few years. This structural impediment will delay the adoption of internet radio. From my perspective you can delay technology adoption but you can not prevent it, at least until now.

Everyone can agree that the artists need to be compensated for their efforts. However, the internet is just another medium of audio transmission. Unfortunately there are different royalty rates applied to terrestrial, satellite and internet transmission. Of these 3 types of transmission the royalty rates for internet transmission are by far the highest.

Borrowing from the broader internet effort on "Net Neutrality" the internet radio inustry should be advocating for "Broadcast Neutrality". The same royalty rates should apply to streaming, satellite and terrestrial radio. One reason is that the internet provided "no promotional value" for CD sales while terrestrial radio did.

I find it odd that the RIAA finds no promotional value in internet radio yet can not only find promotional value in terrestrial radio but also have to pay fines for payola. Many bands have their own MySpace page that stream a sampling of their content to those who visit their page. All done for promotional value.

While it is true that the CRB should not set rates to ensure the success of a poor business model they should also look at other mediums of content delivery for compensation. By setting the streaming rates at such a high level compared to transmission of terrestrial broadcasts they have favored terrestrial broadcasting over other modes of delivery.

If the royalty rates were normalized across the broadcast mediums we would see a faster transition to internet radio and newer business models for getting music out to the masses. It would be a shame if Pandora or Yahoo's Launchcast were shut down while they are just starting to get going.


Friday, October 13, 2006

On Demand Software Scrum

Somehow I missed this article when it came out in Business Week. It is really about how many companies are looking to deliver their solutions via the internet with Microsoft, Oracle and SAP leading the way.

Amazon's Utility Infrastructure is now complete

Amazon's CEO, Jeff Bezos, spoke at the recent MIT Emerging Technologies Conference. In his talk he unveiled their new EC2 initiative. This complements their S3 Web Services and platform and rounds out a complete compute infrastructure for utility computing.

To me the pricing is quite attractive, $.10/CPU/hour, $.20/GB bandwidth, $.15/GB storage billable monthly. It is still a little bit more expensive than a dedicated server that you could purchase from ServerBeach, but that is assuming you run the ServerBeach server at 100% utilization and use all of the allocated bandwidth each month.

However, when you consider that most machines are run at around 10% capacity then the Amazon Pricing becomes much more attractive. The only trouble I have is finding out how to normalize the CPU pricing. Bandwidth and diskspace have a consistent meaning, CPU doesn't.

For Utility Computing to take off CPU unit pricing will need to be normalized so that it could be treated more as a commodity.

Thursday, September 07, 2006

The MediaGrid and standardization, Join In!

On July 30 I presented at the MediaGrid Summit that was held in conjunction with Siggraph 2006 in Boston. The MediaGrid is a public utility for digital media and is based on new and emerging distributed computing grid technologies. As an on-demand public computing utility, a range of software programs and Web sites can use the Media Grid for delivery and storage of rich media content, media processing, and computing power.

The MediaGrid is actively seeking input to the standards. Please join in the definition of the next generation of public utility!

Wednesday, July 05, 2006

Waffling on terms

David Margulius recently wrote for Infoworld that the Grid is like a Waffle Iron. Nice when used but not used often enough.

His use of the term Grid refers to ganging processors together for use to solve a problem. According to the 451 Group, whom he cites in the article, ISVs are slow to move to this architecture. My sense is that the ISVs need to work an appropriate business model to enable movement to a grid architecture.

Thursday, June 15, 2006

Defining terms, Is Utility Computing SaaS?

Many in the industry use the terms Utility Computing, Software as a Service, Application Service Provider and Grid interchangably. This can lead to a bit of confusion. Let me provide my definition:

  • Utility Computing - The provisioning of compute resources as a utility. Sun Microsystems $1/CPU/hour effort is an example of this.

  • Software as a Service (SaaS) - The provisioning of an application as a service to a customer. SaaS can be a utility but a utility can not be SaaS. ConstantContact is an example of an SaaS company. In this instance, ConstantContact provides both the compute resources and the application.

  • Application Service Provider - This term is out of favor and not used as much anymore. Companies in this space were more Utility Computing providers. Exodus is an example of an ASP. They sold "outsourcing" of your applications to their datacenter. They defined the fee they charged you in Utility Computing terms, i.e. technical.

  • Grid Computing - The ganging of processors together to accomplish compute intensive tasks. Sometime this also refers to ganging more than just the processors together and includes storage.
Distinction between these terms is critical to sorting the players and watching the industry mature.

Saturday, June 10, 2006

The Sisyphus Problem - The Net Defect

Meeting the quarterly revenue goals selling computer hardware is a Sisyphean task in itself. The technology trends associated with computer system development makes the task more difficult.

Selling a certain amount of computer equipment represents a certain amount of CPU, memory, disk and in some cases bandwidth. Given that cost of all of these components is decreasing makes it more difficult to acheive flat quarterly revenues. IDC reported in their February 2006 Worldwide Quarterly Server Tracker:
Volume systems grew 7.3% year over year and the segment continue to be the catalyst for growth for the server market overall, gaining favor with SMB and enterprise customers alike. After four consecutive quarterly increases, revenue for midrange enterprise servers declined 11.5% year over year and the high-end enterprise server market showed a 1.7% decline year over year, the fifth consecutive quarter of declining revenue for high-end enterprise servers.
Basically, customers are buying more processing, storage, memory and bandwidth than ever, yet traditional hardware manufacturer's revenues are in a tail-spin. The question is how can these manufacturers take advantage of the technology trends instead of being a slave to them.

Friday, June 02, 2006

Software as a Service Myths?

BusinessWeek has an interesting article on Software as a Service. The article explores 7 myths about Utility Computing. One of the Myths is:

SaaS is just another version of the failed application service provider, or ASP, and hosting models of the past, and will suffer the same fate as its predecessors.

I agree with the author, this is a myth. Though my reasons are a bit different. ASP's had trouble in the past because:
  • Technology was not mature enough to deliver applications as utilities.

  • Pricing models hid the value of the utility.

  • ASPs sold against their natural partners.

  • ASP solutions are technically oriented and horizontally priced.
Contrast this with Internet technology is mature enough to provide their software as a service. Unlike traditional ASPs, they are not selling bits and bytes, their pricing model connects to the users in their terms . Their solution is business focused and vertically priced, the price connects directly to the benefit the customer derives in using the service.

There is a distinction between providing compute resources like a traditional ASP and providing Software as a Service. Similar to the distinction between electricity as a utility and a toaster being a toasting utility. It is this distinction horizontal vs vertical solution orientation that really differentiates the SaaS models from the ASP models.

Thursday, June 01, 2006

Technology "Laws" driving the shift

There are a number of technology laws driving the shift in the economy from atoms to bits. Among these are Moore's Law, Metcalfe's Law and Gilder's Law. These three laws together drive the industry faster than any one of them on their own and were described at Sun Microsystems as the Net Effect. Graphically it would look like this:

The question is where will the processing be done? One perspective is that the net will do the processing. John McCarthy, noted Stanford computer science professor said in 1961:
"If computers of the kind I have advocated become the computers of the future, then computing may someday be organized as a public utility just as the telephone system is a public utility... The computer utility could become the basis of a new and important industry."
The shift as envisioned by McCarthy hasn't happened yet. I believe that McCarthy's vision of utility computing will enable Negroponte's transition from Atoms to Bits. In some respects this is happening. I will write more on this in upcoming posts.

Saturday, May 20, 2006

Atoms to Bits

In 1995 Nicholas Negroponte published Being Digital. The main premise of the book is the shift in our economy from one based upon Atoms to one based upon Bits.

He starts the book with a story about being asked to declare the value of his laptop computer to the security guard at a major corporation. He said "Roughly, between $1M and $2M". The security guard said that could not be true and placed a value on the property pass of $2,000. Of course this represents the value of the computer but not the value of the information that resides on the computer.

Fast forward to today. The shift from Atoms to Bits is transforming business models. The winners will be those companies that see this transformation and implement the infrastructure and business models that moves customers in this direction. A few examples are newspapers like the Boston Globe and the New York Times, developing pictures using Shutterfly and buying music from the iTunes Music Store.

In future entries I will discuss the trends driving this transformation and the business models and products already moving from atoms to bits.

Thursday, May 18, 2006

Software as a Service (SaaS) conference

Software as a Service is one form of Utility computing. The delivery of software or more specifically a solution to a customer hosted off-site. IDC's broad image of this space looks like this:

Souce: International Data Corporation

Examples of solutions in this space are, sugarCRM, Shutterfly and many others. This application space is moving quickly and trade shows dedicated to this space are starting to crop up. One such show is the Software Business Transformation Summit in Las Vegas, NV on June 13 and 14. It will be interesting to see how the definition of the applications evolve compared to the "public utility" used to deliver those applications.

Saturday, May 13, 2006

A major shift in the industry's fundamental economics

I have been meaning to start a blog on the business aspects of Utility Computing. This is an area where I have focused my efforts in the last few years of my career, a shift in the way people will buy applications and computing. A recent Fortune FastForward article headline has prodded me to begin. It states:

Microsoft's cash versus Google
The software giant's plan to build datacenters the size of 10 Costcos, complete with electrical substations, signals a major shift in the industry's fundamental economics

I believe that the shift in this direction happened many years ago for those so inclined to look for these type of shifts.

First a little background on myself. I have spent the last 11 years working at a wonderful company, Sun Microsystems, led by a truly visionary leader, Scott McNealy. Scott has been evangelizing the future of computing starting with Sun's tagline "The Network is the Computer" and pushing the "big friggin webtone switch".

The trouble for Sun is that it is hard to convert a business model built around selling hardware to one built around selling computing as a utility. Sun's move into Utility Computing, selling CPU cycles at $1/CPU hour, is a step in the right direction placing Sun in the position to become a public utility for CPU cycles.

It seems that more and more people are starting to understand the shift in the industry's economics. There are a lot of efforts moving the industry further in the direction of Utility Computing. Going forward I will write about these efforts and how utility computing will impact us all.