The business
potential and operational efficiencies of using new Multiprotocol
Label Switching (MPLS) networks to carry IP traffic are quickly
becoming apparent. It is only a matter of time until IP/MPLS becomes
the industry’s multiservice network technology platform of choice.
However, incumbent service providers can’t ignore the fact that ATM,
Frame Relay, and private-line networks are hard at work today
generating the majority of their data service revenue.
For example, more
than 35,000 enterprises in the U.S. alone subscribe to Frame Relay
access services, according to research firm Vertical Systems Group.
In some carrier networks, this traffic is transported across backbone
WANs constructed purely of Frame Relay switches. More often, Frame
Relay subscriber traffic is aggregated with other traffic streams and
is transported across a high-capacity ATM backbone. In fact, ATM is
thriving as the dominant architecture in carrier backbones in most
service providers’ core data networks. As such, most of the available
expertise in WAN installation, configuration, and troubleshooting lies
in the areas of ATM and Frame Relay.
Because of their
substantial investments in expertise and capital, it is impractical
and economically unfeasible for service providers to simply toss out
their existing networks and replace them with IP/MPLS-specific
infrastructures. So how do they resolve the dilemma of protecting
their current businesses while preparing for an IP/MPLS future? One
way is to slowly begin introducing telephony-grade, multiservice
switches into their networks. These switches should integrate support
for legacy networks, IP/MPLS, and all associated management and
support systems.
Today’s Switching
Landscape
ATM, Frame Relay,
and private-line networks are mature and have been working well for
carrying corporate traffic. ATM is particularly flexible, supporting
interface speeds to OC-48c/STM-16 and offering inherent class of
service (CoS) capabilities for optimizing the performance of various
traffic streams, such as voice, video, LAN data, and SNA.
However, these
switches are not fail-proof, as evidenced by several well-publicized
system-wide network outages in recent months. For this important
reason, the multiservice switches in place today will eventually
become obsolete, at least in certain network segments.
ATM switches, for
example, were designed in the mid-1990s and have already reached the
status of a legacy technology. Many of today’s ATM switches can be
upgraded with MPLS software for delivering IP services. However, these
platforms do not have the reliability levels necessary to carry
tomorrow’s huge mix of mission-critical multimedia IP services. This
is largely because they are based on shared software architectures in
which the failure of a single process affects other processes.
High-availability switch architectures are of particular importance at
the edge of the network, where subscriber traffic is aggregated.
Because customers usually link to an edge access device with a single
connection (rather than with redundant mesh links, which are prominent
in the backbone), the system uptime of edge switches plays a large
role in a service provider’s ability to deliver high customer service
levels.
Next
page >>
Page 1 of 7
