Ericsson (Nasdaq: ERICD) is planning to launch a novel DSLAM at the ITU Telecom Asia 2002 trade show in Hong Kong next week, in a move aimed at securing a position in the emerging market for Ethernet access equipment.

Ericsson's DSLAM (DSL access multiplexer) is novel because it's so small. It's not much bigger than a cellphone and only handles 10 DSL lines. But lots of them can be packed directly into a telecom operator's main distribution frame (MDF), the piece of equipment where subscriber lines are terminated.

The result is a DSLAM that can scale from a mere 10 lines to a whopping 4,000 lines per 7-ft cabinet, according to Peter Linder, technical director of Ericsson's broadband access business unit.

The unconventional design -- traditional DSLAMs are either honking big chassis or 24- or 48-port pizza box devices -- makes it economic for operators to deploy DSL in places with fewer customers. A 10-line entry-level installation costs a mere $2,500, everything included, according to Ericsson.

At the other end of the scale, Linder says the 4,000 lines per 7-ft cabinet is "twice the density of anything else on the market". The ability to add capacity in small increments also means utilization levels can be kept high. Operators don't have to buy gear until they've got customers to use it. The bottom line is that costs can be kept to between $10 and $20 a port, compared to about $40 for traditional DSLAMs, according to Ericsson.

Ericsson's DSLAM modules are also unconventional - although not unique - in being equipped with a Gigabit Ethernet uplink to the broadband remote access server (BRAS) rather than the conventional ATM uplink. Linder says this helps eliminate a bottleneck because most ATM connections are STM1 (155 Mbit/s). He also notes that this bottleneck is becoming a big problem as broadband subscribers use higher bandwidth applications.

Pursuading incumbent carriers to go with Ericsson's Ethernet approach might be difficult, however, judging by comments made by a network architect at one such operator. Speaking anonymously, he told Light Reading he wouldn't consider Ethernet for a couple of reasons. First, it wouldn't fit in with all of his existing infrastructure, which was ATM. Second, he valued ATM's quality of service guarantees. The idea of Ethernet compensating for this by simply throwing bandwidth at the problems wouldn't work, he said. There would still be jitter problems, and curing them (the goal of the Ethernet in the First Mile Alliance (EFMA)) could make Ethernet just as costly as ATM. He also said that modern DSLAMs have higher capacity ATM uplinks - STM 4 (622 Mbit/s) and even STM16 (2.5 Gbit/s) in at least one case.

Ericsson's approach is more likely to find favor with new carriers deploying metro Ethernet technology from scratch, according to Michael Philpott, broadband communications analyst with Ovum Ltd.. This explains why Ericsson is focusing its marketing efforts on Asia and Europe, where metro Ethernet technologies have won wider acceptance than in the U.S.

Philpott thinks Ericsson has a first with its modular DSLAM design and likes the concept. However, he wonders whether incumbent operators would be comfortable installing Ericsson's DSLAM modules directly in their MDFs. There may be reasons why they wouldn't like it, he thinks.

Philpott also has reservations about another ostensibly nifty aspect of Ericsson's DSLAM modules - that individual subscriber lines can be configured as ADSL or S-ADSL (symmetrical ADSL), whether they're POTS or ISDN lines. This helps ensure high levels of utlization and also makes it easy for operators to upsell services, offering S-ADSL to ADSL subscribers who want higher upstream bandwidths.

"It looks like a good idea, but can you configure different SLAs (service level agreements?" asks Philpott. He's leery about using the same equipment to offer ADSL to home users and S-ADSL to business users because ADSL has garnered a reputation for poor reliability.

— Peter Heywood, Founding Editor, Light Reading
www.lightreading.com