New Contender in Networked Storage
A company has developed a new technology that promises lower costs in the popular networked-storage market. Coraid's ATA (advanced technology attachment) over Ethernet (AoE) will now compete with technologies such as Fibre Channel over IP and iSCSI (Internet small computer system interface).
Networked storage is popular because it makes it easier for companies to share storage among multiple servers, add capacity, and centralize system management. However, the technology typically has been expensive.
AoE, on the other hand, uses technologies that lower networked-storage costs, explained Coraid CEO Jim Kemp. The technology's specification is openly available on the company's Web site (www.coraid.com/documents/AoEr10.pdf).
The approach lets users access information from or write data to standard PC storage drives—ranging from one to an array of disks—via Ethernet networks.
Coraid's ATA over Ethernet technology promises lower costs for networked storage. When a user wants to access data from or write data to networked storage, an AoE initiator broadcasts a message to discover the available storage devices on the network. Available targets respond with their addresses. The initiator then sends them the messages and commands for the desired activity, which the devices execute. ATA serves as the electronic interface between the user's computer and the disks. It integrates the controller on the disk drive. This simplifies the disk-access interface. It also places basic functions, including self testing and sector remapping, on the disk and thus doesn't use the host system's resources for these activities.
AoE will offer maximum data rates that are the same as that of the fastest Ethernet technology, Gigabit Ethernet, which runs at 10 gigabits per second. iSCSI, which also uses Ethernet technology, provides the same rate, while Fibre Channel generally operates at up to 1 gigabit per second.
The Internet Engineering Task Force's iSCSI standard links distributed data-storage facilities by carrying SCSI commands over IP networks, including those based on Ethernet.
Fibre Channel over IP translates Fibre Channel control codes and data into IP packets for transmission via optical networks between geographically distant storage area networks. Both iSCSI and Fibre Channel use SCSI drives.
AoE offers low costs and simpler operations because it works with standard, mass-produced hard drives, rather than the more expensive SCSI drives, explained Kemp. It also uses common Ethernet equipment, rather than the less common and pricier Fibre Channel networking hardware, he said.
In addition, he noted, Fibre Channel networks are complex and include many functions that aren't part of Ethernet networks. Buying equipment and training staff to work with Fibre Channel are, therefore, expensive.
Unlike iSCSI and Fibre Channel, AoE doesn't work with TCP/IP and thus is limited to use only with locally networked storage.
However, Kemp noted, not using TCP/IP also reduces the number of processing and connection issues, and eliminates the need to manage IP addresses. This makes AoE faster and simpler.
AoE's best uses include centralized data storage for companies—such as those that store digital photos, or medical or academic-research records—that have found other technologies too expensive.
Competing in the networked-storage market is attractive because it is worth about $30 billion, said Robert Gray, research vice president for storage systems with market-research firm IDC.
However, he doubted AoE's value, saying the networked-storage market is already crowded with better-established and equally effective technologies.
Coraid's AoE products include EtherDrive storage appliances and the Linux NAS Gateway CLN20, which links servers and PCs with network-attached storage.
Kemp said Coraid plans to seek IEEE adoption of AoE as a standard in the future.
HP Releases Computer on a Sticker
Hewlett-Packard has developed an inexpensive, rewritable storage device with computing-and radio-based communications capabilities that can be placed on a sticker the size of half a rice grain.
Users could stick the Memory Spot on any number of objects—such as product packages, hospital wristbands, and walls near displays in art or history museums—so that users with data readers could access information about them, noted Howard Taub, a company vice president and associate director of HP Labs.
HP says the Memory Spot would be more useful than radio frequency identification (RFID) tags for these purposes because it would be less expensive; hold more data, including video and audio; and offer more computing functionality.
Because of the 2.45-GHz frequency range in which it operates, Memory Spot can transfer data at up to 10 megabits per second, compared to RFID's maximum 384 kilobits per second.
Given its higher data rate, HP built the Memory Spot to hold between 256 kilobits and 4 megabits of flash memory, compared to RFID's data capacity of between 128 bits and 256 kilobits. Memory Spot thus has enough capacity to store relatively large amounts of information.
An RFID device costs between 15 cents for a basic tag to $30 for a high-end implementation like the one Boeing is building to identify its aircraft parts. HP says it hopes a Memory Spot will cost about $5 initially.
A prototype Memory Spot uses a 1.4-square-millimeter chip with integrated NAND flash memory, processor, capacitor, and antenna as well as a radio. Manufacturers must add antennas to RFID chips, which increases their cost, said Taub.
Users can write, erase, rewrite, and read data using Memory Spot chips, which have a small file structure and can handle authentication and encryption.
To access information, data readers must either touch or get within 1 millimeter of a Memory Spot. This makes it unlikely that a hacker could intercept transmitted material, according to Taub. Thus, he said, Memory Spots might be appropriate for use with driver's licenses, credit cards, or passports.
For the Memory Spot to succeed in the marketplace, HP will have to convince cell phone, PC, and other device makers to include readers in their devices.
Moreover, HP will have to identify the applications that are best suited for the Memory Spot, said Carl Howe, a principal at Blackfriars Communications, a market research firm. "I'm not sure I'm completely clear on what to use it for," he said.
HP says it could either make and sell Memory Spots itself or license the technology to others. According to Taub, the company isn't sure when it will release the Memory Spot commercially but plans to conduct a demonstration project this year.
Making Web Science an Academic Field
A group of university researchers, including Internet pioneer and MIT senior researcher Tim Berners-Lee, is taking the first steps in developing a new academic field of study called Web science.
Researchers at MIT and the UK's University of Southampton have formed the Web Science Research Initiative (www.webscience.org). They have expressed hope that universities worldwide will adopt the new field of study.
Proponents say the Web has dimensions that extend well beyond traditional computer science to include emerging research in social networks and the social sciences on how people behave online. The WSRI will thus work on a multidisciplinary approach to understanding how people behave and interact online and to studying the social and technological implications of growing Web adoption.
"As we move into the next generation of the Web—the Semantic Web or Web 3.0—it will spawn lots more businesses and community activity," said Wendy Hall, Southampton University professor and head of its School of Electronics and Computer Science. "The interesting thing is trying to predict what might happen and study how the Web does and will evolve. This is the heart of Web science."
Web science research could lead to more effective Web applications, systems, and businesses that better reflect the social relationships they entail, stated Hall.
The Web's ongoing importance necessitates the teaching of a very different kind of computer science, according to University of Maryland professor Ben Shneiderman. "The Web presents new opportunities," he explained. "There are hundreds of millions of uses for any system and new applications, whether you're talking about e-commerce or education. That's the excitement and the opportunity I hope we can embrace."
Understanding Web-related concepts such as universal usability and human-computer interaction will be important for future technologists, but current academic programs don't combine social and computer science approaches to fully address these issues, he explained.
"All this doesn't mean we won't need computer scientists or engineers in the future," Hall stated. "But we also need a new type of person who understands the technology and human behavior and the way society works."
Advocates are just beginning to plan what a Web sciences curriculum might include. Many of them attended a conference on the topic at the University of Maryland in December. Southampton University has begun supervising PhD candidates in this area and is planning to offer masters and undergraduate programs.
The most likely approach will be adapting existing classes to include more Web-relevant issues rather than adding new classes or adopting a wholesale curriculum reform, Shneiderman predicted. He expressed hope that computer science departments will be receptive. If not, he said, "They will not survive."
These changes won't be easy because they already face resistance from faculty members who don't want change or see no need for it, he noted.
Google and IBM are supporting the WSRI effort. In addition, proponents are talking to government agencies throughout the world—including the US National Science Foundation and the UK's Engineering and Physical Sciences Research Council—that finance scientific research.
The WSRI plans to hold Web-science workshops, to be announced on the group's Web site, and get input from various sources before trying to attract government, foundation, and corporate funding, Hall said.
"We are in the startup phase of the initiative," explained Hall, "so we have a long way to go."
News Briefs written by
Linda Dailey Paulson, a freelance technology writer based in Ventura, California. Contact her at ldpaulson@yahoo.com.
DOWNLOAD PDF 
