Research could challenge existing notions of the network layer hierarchy.
Now that information service providers are rushing to provide global high-speed connectivity, differences of opinion have emerged about which type of protocol will best link diverse users. The potential for a dramatic shift in network methodology and the inability to predict what the future communications protocol of choice will be have left developers in a quandary.
The debate centers around the differences between two alternatives. When asynchronous transfer mode (ATM) was introduced, it was heralded as the communications protocol that would offer “network nirvana,” according to one industry professional. Now, one camp in the multisided debate is proposing to bypass ATM and rely on Internet protocol (IP).
Discussions about the use of Internet protocol and asynchronous transfer mode technology in network communications have reached an almost religious intensity in the data and telecommunications industry. Experts agree there is a need for a highly capable mechanism to transport voice, data and video communications through networks and over the Internet. But how the process is facilitated has become a bone of contention between many service providers and equipment vendors.
In the discourse between the “bellheads” and the “netheads,” as they are being called, one side asserts that the two technologies work together, and should in the future, given ATM’s proven ability to handle multimedia in addition to voice and data. To ensure the quality of service offered, especially for voice communications, providers maintain that ATM still prevails. But researchers are looking at alternatives to building costly ATM infrastructure, and Internet protocol has captured their attention.
Many people agree that IP will be used in the future, but whether it will be used without ATM remains to be seen. Most agree that if IP will be used apart from ATM, it will need to undergo some sweeping changes to provide the level of service for which ATM is recognized.
Both IP and ATM reside on the open system interconnection-reference model (OSI-RM). This International Standards Organization model illustrates communications system architecture and the operational processes within the system. The chart consists of seven layers in a hierarchy that defines the services provided at each level. Each layer relies on subsequent and previous layers to perform a series of functions that together provide effective communications transmission.
ATM is a communications protocol that is housed on the OSI-RM layer two, or datalink layer. This layer operates above the lowest layer, called the physical layer, and sets up the datalink to transfer data in a network. Flow control, congestion management and error detection in the physical layer are also handled at this level.
Designed for a mixed environment to handle voice, data and video communications without sacrificing quality, ATM has 53-byte fixed-length cells that contain the information housed in the mode’s structure. Information is placed into these cells and transported across a network in a connection-oriented environment using a virtual connection or path. The cells travel together over the network and arrive in the order in which they were sent. The fixed length of the cells reduces processing time.
Because of its asynchronous nature, ATM information is sent in bursts as opposed to a continuous, or synchronous, transmission. This allows the technology to be implemented when needed—the popular capability of bandwidth on demand—to scale up and down capabilities without the unnecessary use of network resources.
IP emerged from the growth of the Internet and is an application that rides on an ATM structure. It operates at the OSI-RM layer three. This network layer provides the capabilities to transfer data from a source to a destination through a network or among networks. Routing, flow control and error control functions are also performed at the network layer.
Unlike ATM, IP uses packets to hold and transmit information. Packets vary in length and operate in a connectionless environment, where they separate from each other, go through different routes to reach their destination, and reassemble after reaching their intended end point. The service is categorized as unreliable because packets can get lost or can be delivered out of order; it is considered a best-effort service.
The debate over the use of IP without ATM centers on the capabilities of the two technologies. The issue of service quality tops the list of priorities. How the protocols can effectively be used to achieve a high level of quality for such applications as voice is a pressing question. Experts agree that ATM allows the most acceptable quality of service for voice transfer over the public Internet. With IP, which was designed primarily for data communications, there is an inherent possibility of quality downgrade when used for voice transmissions in networks that are not optimally designed.
“Voice can be passed over IP, but what has to be done to make that work would produce a noticeable difference in quality,” Tom Nolle says. He is president of CIMI Corporation, a Voorhees, New Jersey, company that specializes in network integration and market research. “We’re not going to be satisfied by a best-efforts guarantee,” he adds.
L. David Passmore, founder and research director of NetReference Incorporated, Sterling, Virginia, a network architecture consulting firm, agrees that quality of service for voice using IP has not yet arrived. Until Internet protocol can be developed, he maintains, industry has little choice but to develop and use ATM products.
Because demand for quality voice service far exceeds demand for data transmissions, Nolle believes that existing voice-centric networks will determine which protocol will dominate in the industry. “The road to the future is going to be measured in dollars,” he offers. Many experts agree that until IP can offer the same quality of service for voice, it cannot be effectively used as a primary protocol without ATM.
Also at hand are the issues of performance and traffic engineering. ATM traffic flows ideally through switches, while IP traffic is managed mainly through routers. As information works its way across a network in a connectionless environment, every router determines where to send traffic. However, the router does not take congestion into account and can opt to send packets through a short, congested path instead of a longer, uncongested one. ATM’s fixed path method monitors traffic and allows for more network control.
Now, developers are trying to enhance IP’s capabilities by retrofitting quality of service into the protocol. One solution—differentiated services—promises to adjust for quality issues by determining the level of service quality desired and fulfilling that need accordingly. “We’re not there yet,” Passmore acknowledges about IP efforts. The technology is still immature, but it is coming, he offers. He also believes that items such as terabit routers will help the performance issue associated with discrepancies between IP and ATM.
Still, the next-generation Internet protocol, or IPng, also referred to as IPv6, is being designed to address the quality of service issue. Developers want to make IP a connection-oriented service to handle multimedia and other synchronous data.
However, some experts argue that in changing IP to provide ATM’s quality of service level, technologists are reinventing asynchronous transfer mode technology. They ask why developers are trying to create something that already exists. One industry professional suggested that it is a matter of vendors trying to capitalize on each new possibility.
Many people want to cut out the middleman and have IP do what ATM does, but Nolle says the two concepts go hand in hand. He believes IP will be the dominant new service type with a role for ATM in the infrastructure.
According to Passmore, it is a battle of the “netheads” versus the “bellheads,” with the netheads representing the IP-centric community and the bellheads supporting the role of ATM. Many traditional carriers and big service providers are on one side holding fast to the notion of ATM in conjunction with IP, while many equipment vendors and newer telecommunications companies are looking toward IP. Some vendors that produce equipment for both sides clearly take the middle-of-the-road position.
While industry professionals are leading the fight from their marketing offices, standards organizations maintain there is no contest. The ATM Forum and the Internet Engineering Task Force (IETF) are organizations that promote network technologies and standards. The ATM Forum discusses issues and determines policy related to ATM technology, while IETF focuses on issues related to the Internet and its protocol. The two organizations often include corporate and individual members from both communities, and they cooperate to make sure there is no duplication of efforts.
IETF Chairman Fred Baker maintains that the two options do not compete because they are different communications layers on the OSI model. He contends that the two work together, stating that approximately 75 percent of the Internet backbone traffic crosses an ATM infrastructure at some point during transmission.
Baker says there is no technical quarrel, but a dialog has arisen that stems from IP-specific companies versus their ATM-specific counterparts. “The only fight is in the minds of marketing folks from companies that specialize in one such technology or the other, or from conference-planning organizations and the members of the media who have gotten lost in the hype,” Baker summarizes.
Members of the ATM Forum also state that there is no contest between IP and ATM. “They are both going to evolve,” says ATM Forum Secretary Rick Townsend. George Dobrowski asserts, “You won’t see one versus the other. It’s going to be the two.” Dobrowski is president of the ATM Forum and a member of the organization’s board of directors.
The ATM Forum must ensure that IP can be carried over ATM and that certain equipment will interoperate on a carrier network, ATM Forum Technical Committee Vice Chair Bob Klessig says. Klessig argues that the issue is not “us versus them,” but rather, it is an issue of IP and ATM complementing each other.
Dobrowski refers to the OSI model in addressing the IP/ATM question, “Those layers are there for valid reasons. If you want reliable transport, you cannot ignore the layers,” Dobrowski says. He believes, along with many in the standards organizations, that the discussion should be about how the two protocols are going to evolve.