A viable choice for low-speed ATM Carriers may need to bundle their T-1 ATM lines for service differentiation and customer flexibility

Asynchronous transfer mode has gone through many phases in its relatively short life, ranging from a super high-speed transmission technology to a service purely driven by video and multimedia applications. Today, ATM is a real solution for customers because it has been grounded with low-speed services, equipment and specifications, like the ATM Forum's ATM inverse multiplexing spec that is nearing completion.

The industry has heeded the requests of its customers with equipment and services that support T-1 ATM. Although only a few service providers are offering T-1 ATM services, it is expected to continue to grow in popularity.

To help facilitate that growth, the ATM Forum has drafted a number of specifications to address low-speed ATM. These specs will offer users flexibility and a migration path to higher speeds. One specification, ATM inverse multiplexing, will let service providers provision sub T-3 rates to offer customers a solution for today and a migration path for tomorrow.

"ATM inverse multiplexing gives you the granularity to ramp up to T-3 and then to OC-3. That is the scalability and flexibility that ATM has promised all along," says Tim Burke, communications analyst at The Yankee Group. "This is what ATM is supposed to be all about."

The Pending Spec ATM inverse multiplexing is the answer to customers' need for interim speeds between T-1 (1.544) and T-3 (45 Mb/s) ATM services. The pending specification will give equipment manufacturers the guidelines necessary to design ATM switch upgrades and new products that will allow service providers to bundle multiple T-1 lines. Based on ATM inverse multiplexing, service providers will offer fractional T-3 ATM services giving their customers the sub T-3 speeds they need.

Service providers can provision multiple T-1 ATM lines to a customer location from an edge switch or an ATM concentrator. Bundled T-1 lines will give users a seamless chunk of bandwidth; three T-1s will offer customers 4.632 Mb/s worth of bandwidth, for example.

The process involves ATM cells traveling from the ATM or the cell layer where the voice, video or data traffic is put into cells, to the physical layer where the cells are divided for transmission over multiple T-1 links. The cells travel separately along their assigned T-1 line and are reassembled at the other end of the network at the physical layer. They are then sent to the ATM layer as if the traffic were transmitted over an ordinary connection.

ATM inverse multiplexing has been a work in progress in the ATM Forum since late 1994, says Rick Townsend, chair of the physical layer group at the ATM Forum and distinguished member of technical staff at Lucent Technologies, Murray Hill, N.J.

Many issues regarding the pending specification have already been worked out, he says. The latest and possibly final issue, he says, is how to inverse multiplex non-synchronous T-1 lines.

The ATM Forum is now evaluating the non-synchronous portion of the specification, he says. Once that is resolved, additional work maybe done in other committees such as the signaling and traffic management groups. Townsend expects the specification will be ready for a straw vote to all of the principal members at the next meeting, in June or August.

The Carrier Market Impact Carriers see ATM inverse multiplexing as a way to differentiate their ATM service offerings. AT&T has been offering T-1 ATM service since the end of last year, and it plans to bundle ATM inverse multiplexing support with its InterSpan services based on customer feedback.

"The ATM inverse multiplexing feature is crucial to our customers," says Paul Moravek, product manager for ATM. "T-1 doesn't have enough throughput or bandwidth for many of our customers, but they don't need 45 Mb/s or the cost. Many users need 4.5 to 6 Mb/s of bandwidth."

AT&T is using GlobeView-2000, Cascade's B-STDX and StrataCom's IPX switches throughout its network to provide ATM services. AT&T will incorporate ATM inverse multiplexing upgrades by some of these vendors to support new services.

Sprint also offers T-1 ATM services and is poised to move toward ATM inverse multiplexing support. "ATM inverse multiplexing is an enabling technology," says Kathy Gadecki, group manager of ATM service. "We have customers that have a requirement for greater than T-1 bandwidth, but are not ready for T-3."

Sprint-like other service providers-initially offered T-3 as its lowest ATM speed. When ATM first made its way into service providers' networks it did under the guise of a very high-speed service, supporting transmission speeds from 45 to 155 Mb/s. Because of this, Sprint is supporting more T-3 ATM connections to date, but the IXC expects to see a lot more demand for T-1 ATM as the service becomes more widely available.

LDDS WorldCom is backing another service for its customers' lower speed and lower bandwidth requirements. The carrier recently announced high-speed frame relay service that supports transmission rates up to 45 Mb/s. The service will run on the IXC's Data Concourse broadband network that also supports ATM traffic and frame relay-to-ATM service interworking.

Purely from a data standpoint, high-speed frame relay makes sense for some users, but it does not support as many traffic mediums as ATM, which is designed with the intent to support voice video and data traffic simultaneously. Recently, though, products that support voice over frame relay have hit the market and work well for many users.

Higher speed frame relay will let users keep the same equipment, with the exception of a possible upgrade to handle higher speeds, and offers users a familiar, proven technology, says Rick Malone, analyst at Vertical Systems Group, Dedham, Mass.

LDDS WorldCom is backing high-speed frame relay because many applications designed to run over ATM- such as large file transfers, boardroom quality videoconferencing or virtual reality-need the 45 to 155 Mb/s worth of ATM bandwidth, says David Nathao, manager of data product marketing at LDDS WorldCom.

To provide its customer base with choices, LDDS WorldCom plans to roll out T-1 ATM services for low-speed access, but sees its high-speed frame-relay service fitting the bill for many of its customers. "We will probably support [ATM inverse multiplexing]," Nathao says.

Analysts say existing frame relay customers are more likely to upgrade their service to higher speeds than to jump to a different service entirely. For example, if a company has an application with T-1 bandwidth requirements it will probably bring it up on frame relay, says Malone. The same company might have ATM at its headquarters running at T-3 rates and it could use frame relay-to-ATM interworking.

Interworking is a specification from the Frame Relay and ATM Forums' designed to let frame relay frames travel over a cell-based network like ATM without converting the frames to cells or vice versa. This specification, like ATM inverse multiplexing, provides users with a migration path to ATM and to higher speed services. The maturation of these lower speed specifications will support analysts' predictions of increased T-1 provisioning over the next few years.

Vertical Systems Group is predicting a significant increase in low-speed T-1 ports compared with T-3 ports (Figure 1). "We will see a lot of sites running a few T-1s," says Malone. "But T-3 offers a higher speed service and will most likely prevail after 1998 as the choice for many users after T-3 access prices come down ."

Monthly port charges for T-1 ATM services average around $2500 compared with $13,000 for T-3 ATM. Carriers have not yet determined if the pricing structure for ATM inverse multiplexing service will be as straightforward as adding an additional $2500 for each 1.544 Mb/s of bandwidth requested, but the cost savings is apparent.

There is a good chance that T-3 pricing will come down, but no one really knows how fast that will happen. Analysts have mixed views regarding when the gap between T-1 and T-3 prices will close, but they all agree it will eventually happen.

Rates for a T-3 line are going to come down close to T-1 rates in the near future, says Tom Nolle, president at Cimi Corp., a Voorhees, N.J.-based consultancy. When that happens there will not be a need for ATM inverse multiplexing, he says.

Townsend, from the ATM Forum, agrees that if there is no cost savings with fractional T-3 then there is no point in offering the service. But he points out that his crystal ball has yet to predict price reductions or increases in the carrier market. Since the price difference is still significant today, there is a need for ATM speeds between T-1 and T-3, he adds.

One Small Step Behind ATM inverse multiplexing will definitely help in the deployment of ATM services, says Bob Folgem, product manager of fast packet services at BellSouth. A 45 Mb/s service is more than most people need, says Folgem. "This is driving us to look at inverse multiplexing. Customers are looking for a more cost-effective way to meet their bandwidth requirements." BellSouth's customers have identified the need for low-speed ATM, but the LEC may still be far from offering a T-1 ATM service or inverse multiplexing. BellSouth has a request for information out for a small to medium-sized edge switch that is essential in provisioning low-speed and ATM inverse multiplexing services and has narrowed it down to three vendors. The LEC has Fujitsu's Fetex 150 ATM switches installed in its backbone.

Some RHCs would prefer their customers to support ATM inverse multiplexing from their side of the network by installing an ATM aggregator or an access device. But BellSouth says it would prefer an incorporated approach. "If it's native on the ATM switch, then that would be better," he says.

U S West will offer lower speed access to ATM that includes T-1 service, native LAN services and interworking , says Greg Gum, executive director of solutions development at U S West. The verdict is still out for U S West and other LECs regarding ATM inverse multiplexing.

Though U S West has not decided whether it will offer ATM inverse multiplexing, the RHC is poised to offer sub-T-3 rate service, but from the customer site. Based on a relationship U S West has with ADC Kentrox, it can deploy ADC's access concentrators along with its ATM service. The devices will also support the inverse multiplexing feature, says Gum, giving users the flexibility to multiplex T-1 ATM lines from their site.

Pacific Bell is looking to introduce low-speed access to ATM at T-1 rates, says Ed Briones, FasTrak ATM cell relay product manager. "We need to deal with the legacy networks out there now." ATM inverse multiplexing will offer the scalability to support Ethernet, token ring and PBX traffic, he says.

Like U S West, Pacific Bell would prefer ATM inverse multiplexing support to come from the customer site. But if ATM service competition moves into Pacific Bell's backyard, it will rise to the occasion, Briones adds. "We want to be the full service provider," he says.

Looking Toward the Future ATM inverse multiplexing and other lower speed migration technologies such as interworking offer customers the choices they want. For carriers to be successful in new markets or to protect their existing customer base, they will have to offer products and services that will meet this demand head-on.

ATM inverse multiplexing will let carriers bundle their T-1 lines to offer a dynamic, flexible service that will not seem out of reach to its customers in respect to cost or technology.

The hype surrounding ATM as a super high-speed and multimedia driven technology is over. Users want to hear about real services that are going to help them succeed in their businesses.