Author: Jack M. Germain
Some cloud computing vendors, such as 3tera and Nirvani, push their own proprietary platforms and tools, which forces adopters to limit their options and work in a restricted or closed architecture. When these established vendors say cloud, they mean their cloud. As a result, Web developers may believe that, in order to use cloud computing, they must accept limitations in the way they write and build their applications. But that view is a misconception; open standards for cloud computing are already in place and are being
This does not mean that a single cloud computing platform is universally available. But just as some vendors have developed their own proprietary platforms for working in the clouds, so have various open source companies and communities.
“We’re already there. That is the trend I’m seeing,” says Jim Zemlin, executive director of the Linux Foundation. “Most Web-based startups are not buying hardware or software. They are using open source middleware and programming products like Ruby on Rails and Perl.”
What’s in a name?
Cloud computing is more of a process than one set technology. The concept behind what is now referred to as cloud computing has been called a variety of things, including cluster computing, utility computing, grid computing, and on-demand computing.
In its current trappings, the cloud computing model involves distributing computing tasks such as data storage and data center contents to a variety of Internet connections, software, and services accessed over a network. This collection of servers enables users to access supercomputing features. The data is not anchored to one physical location.
The push toward open standards for cloud computing has been going on for some time. This trend toward using open source tools for accessing the clouds is continuing to grow, says Zemlin.
Perhaps the most challenging factor for potential adopters of cloud computing services to consider is which path best meets their needs. According to Zemlin, many organizations are integrating open source products to offer choices for accessing cloud computing service.
For example, a team of developers in the Computer Science Department at the University of California in Santa Barbara recently released the Eucalyptus Project, an open source infrastructure for cloud computing that mimics Amazon’s Elastic Computing Cloud (EC2), under the FreeBSD license. The name Eucalyptus stands for Elastic Utility Computing Architecture for Linking Your Programs To Useful Systems. This software infrastructure implements cloud computing on clusters. Its design supports multiple client-side interfaces. Eucalyptus uses Linux tools and basic Web service technologies.
Not so fast
Some cloud computing players are not disputing the availability of open source products but question how much of a standard exists yet. They are not sure, for instance, how best to apply data management to the cloud.
“We are all for open standards via open source. But there is no clear path yet to what that standard should be,” says Aaron Darcy, director of product line management for the JBoss Division of Red Hat.
Darcy says his company’s top tier enterprise customers are pushing the envelope on cloud computing. They ask for cloudware and want extensions of existing software and standards.
“Our customers want to leverage cloud computing for its economy and quick deployment to market. But they don’t want to reinvent the wheel,” he says.
Red Hat, like other companies, is rushing products to market to fill customer demands in the clouds. For example, Amazon has worked a deal with Red Hat to run some of its open source products, such as JBoss Application Server, in the clouds. That makes sense, Darcy says, because it reflects a natural extension of what his company is doing on the enterprise level.
The goal of cloud computing should be access through open standards, according to Darcy. That is the only way the technology can adjust to new developments without locking in users to an inflexible platform, he believes.
“With cloud computing standards, no one has it right yet, including us. The market is still so young,” he says.
The Linux Foundation’s Zemlin does not dispute the growing pains cloud computing is facing. Clearly, it is at an early stage of development. That means the industry has not yet put all of the needed tools in place.
The industry needs to solve a few things, however, he believes. For example, cloud operators need better manageability applications to allow cloud users to understand utilization rates. Numerous open source tools are doing this now, Zemlin says.
To that end, Zemlin wants to see a consistent way to meter and charge for cloud use. Cloud computing is a game of scale. Its real benefit comes from leveraging economies of scale, he notes.
At this point in the growth of cloud computing, companies that want to take advantage of cloud services have to consider whether to build their own computing clouds or subscribe to others’ cloud servers. These decisions often involve guesswork about which platform or cloud service offers the
most reliabiilty and longevity. This guessing game in part results from the lack of a clearly defined clud standard supported by both proprietary and open source developers.
“Given budget constraints, customers need to evaluate the trade-offs,” says Red Hat’s Darcy. Cost differences are one of these trade-offs. For instance, potential adopters have to weigh the expenses associated with buying into a proprietary cloud platform or a community-sponsored or paid support open source product. Also, moving to the clouds could entail purchassing both hardware and software that could result in additonal upgrades or existing programs or time and money learning to adopt to new programs.
Another trade-off at this stage of the cloud computing game, Darcy says, is concern about security risks and performance hits. Placing a company’s data in somebody else’s cloud configuration, for instance, raises worries about how secure the data is. Even more troubling may be how the additonal middleware layer’s impact on the corporate computers.
“All of the hidden factors are not recognized yet,” Darcy concludes.
A good amount of the growing pains for cloud computing is similar to what the software as a service (SaaS) industry suffered in that technology’s early stages, according to Zemlin. As he sees it, SaaS was a first-generation technology for cloud computing. Soon, cloud computing may have similar benefits for smaller businesses and consumers.
- High Performance Computing
- Internet & WWW