Carrier Channel: ATM Aggregation On Deck

By Khali Henderson

Posted: 8/2003

ATM Aggregation On Deck

By Khali Henderson

BellSouth Interconnection Services
unveiled in July flat-rate pricing for variable bit rate (VBR) permanent virtual
circuits (PVCs) and will follow up in fourth quarter with a complementary ATM
circuit emulation service. Both offers, company executives say, are designed to
allow carriers to aggregate local access traffic more cost-effectively by using
ATM connections.

Jose Blanco, director of product
management for BellSouth Interconnection Services, says the new service
offerings are part of a larger push toward packet-based interconnection with the
company’s local access network.

Even the largest carriers typically
still interconnect to the local access network using TDM special access
facilities and interfaces at DS3 or below, keeping the cost of connecting
lower-speed end user location to an IXC PoP relatively high.

While ATM is slowing making its way
toward the edge (many DSL connections are ATM based today), it shares company
with frame relay-based links. Blanco says BellSouth offers frame relay-to-ATM
interworking, so carriers can aggregate multiple traffic types on an ATM
backbone.

Making this an increasingly
attractive proposition, BellSouth Interconnection Services is offering VBR PVCs
for a flat rate.

Hector Zayas, product manager, says
the new pricing includes a fee to enable the ATM port to accept the PVCs and
charges for the two segments — one facing the carrier and the other facing the
end user — that make up the PVCs.

"That is very different from
the other model where you just pay for the port, the segment charges and a
bandwidth charge based on use," says Zayas, who shared a preliminary
pricing guide showing the flat-rate pricing favors aggregating 30 PVCs or more.
Aggregating 30 VBR nonreal-time (NRT) PVCs at 256kbps over an ATM DS3 costs only
$25 less under the flat rate plan than the regular scheme, but aggregating 100
PVCs is nearly $700 less per month while the cost of aggregating 300 drops by at
least $2,500 per month.

"[The new pricing] facilitates
for the carrier aggregating all kinds of traffic over high-capacity ATM
connections and being able to have a predictable model for cost," says
Blanco. "It scales up with the number of users."

"It’s a carrier-friendly way to
price access into the cloud," he adds.

BellSouth Interconnection Services’
packet cloud consists of packet services areas that are similar to local calling
areas. In some cases, says Blanco, these can be quite large. One runs 130 miles
north and south of Miami, he notes. Blanco says a carrier could access the
service area for the price of an ATM DS1 ($565 per month before volume
discounts) whereas traditional TDM transport for a similar length circuit could
cost upwards of $1,000.

"If you contrast it to buying
leased line transport with mileage [sensitivity], the ATM packet cloud economics
can be compelling. In some cases," he says, "it might make sense to
pay for ATM with circuit emulation to serve a private line application."

Circuit emulation is another product
BellSouth Interconnection Services is grooming for a fall release. The wholesale
carrier will be able to emulate the DS1 private line standard and transport it
over the packet cloud as ATM for applications that require a TDM interface.

Blanco offers the example of a
wireless carrier with a remote cell tower that requires backhaul via T1 because
the radio technology includes a proprietary packet protocol that requires a TDM
interface. "The only way I can serve that to my ATM cloud is to give you a
TDM interface, a DS1 interface," he says, adding that the same wireless
carrier is likely in process of implementing its 3G migration. "Over time,
they are going to need ATM out there, so they start doing more and more business
through the packet cloud and it all supports their strategy of migrating to
3G."

Network-to-Network Interface
Economies of Scale

A BellSouth Interconnection Services
white paper shows TDM access to the local network is relatively high compared to
packet access with optical NNIs.

Consider a frame relay network
example involving the delivery of a DS1 operating at 1.5mbps from an end user
location to an IXC PoP.

On a fully allocated basis, a DS1
frame relay port may cost $100 per month.

Add to this the cost of the TDM
multiplexing equipment required to go from DS3 to DS1 at about $15 per month.

Next add the cost of the local
access itself: If a DS3 with 3/1 mux from the LEC costs $1,000 per month, divide
by 28 to get the prorated DS1 portion and divide again by 0.8 to allow for 80
percent average utilization. The result is about $45 per month.

Thus, getting from the PoP to the
local access network runs about $160 per month.

Now consider a high-capacity ATM
interface, such as an OC12.

An OC12 operates at around 599mbps.
Divided by 1.5mbps, this equates to approximately 390 DS1’s worth of bandwidth.
When engineered for 50 percent oversubscription (390 x 1.5) made possible with
statistically multiplexing packet transport, you get capacity equal to 585 DS1s.
If an ATM OC12 access port, loop and transport from the LEC cost $11,371, then
the prorated portion of a DS1 with 50 percent oversubscription is about $20 per
month.

Now, add the prorated DS1 portion of
an ATM OC12 port’s bandwidth with 50 percent oversubscription at about $2 per
month.

In this example, the difference in
moving from a TDM environment to an ATM environment for access from the PoP to
the local network is $22 compared to $160.