PICMG 2.16 Gains Critical Mass
BY ED BIZARI
It has become clear over the past several years that existing bus-based
architectures for embedded design have been pushed to the utmost of their
abilities. As a result, a flurry of new switching and interconnect
architectures have emerged for standardization.
While used in varying forms over the years, the concept of standardizing
a packet switching backplane architecture took shape with the ratification
of the Ethernet-based PCI Industrial Computer Manufacturers Group (PICMG)
2.16 specification in September of 2001. Since then, PICMG 2.16 has gained
considerable critical mass, with over 30 vendors providing a full range of
2.16-compatible products and numerous equipment manufacturers adopting PICMG
2.16 for next-generation system designs.
The PICMG 2.16 CompactPCI/Packet Switching Backplane standard leverages
the ubiquity of Ethernet to bring a variety of benefits to equipment
manufacturers and, ultimately, carriers and service providers worldwide.
Additionally, several driving forces have pushed, and will continue to push,
the use of Megabit and Gigabit speed Ethernet as a bus connectivity
Ethernet is everywhere, primarily because of its ease of use, global
familiarity, and ongoing ability to continually evolve technically. Since
its standardization by the IEEE in 1983, Ethernet has become the worldï¿½s
most widely used LAN access method. At some point, 95 percent of the worldï¿½s
data travels on Ethernet, and 85 percent of the worldï¿½s installed networks
are Ethernet-based. As a result, Ethernet has become dominant in enterprise
LAN networks and has caused the worldï¿½s most complex network, the public
switched telephone network (PSTN), to begin a migration from a circuit
switching to a packet switching methodology. Overall, this momentum is
pushing Ethernet/IP to become the networking standard of choice, not only in
LAN and telecommunications networks, but also in data communications,
metropolitan area networks (MAN), wireless networks, home networks, and
factory floor networks.
Currently, 100Mb Ethernet is representative of a ï¿½standardï¿½ LAN
connection port, and, according to market research analyst firm Venture
Development Corporation, the worldwide installed base of Gigabit Ethernet
ports is expected to increase more than fivefold by 2004. Additionally, the
10G Ethernet standard, including transmission over SONET (Synchronous
Optical Network) cabling, is nearing completion, targeted mainly as a
carrier technology for metropolitan-area service providers, backbone
aggregation of Gigabit links, or switch-to-switch connectivity.
Recently, Ethernet has made another important leap ï¿½ into the backplane as
a bus connectivity architecture within standards-based industrial computers.
ETHERNET IN THE BACKPLANE
Telecommunications infrastructure compute nodes, such as call processing
servers, multimedia gateways, integrated access devices, or wireless
infrastructure equipment, are based on either proprietary hardware
architectures or standards-based architectures. Depending on the core
competencies of a given equipment manufacturer (i.e., competencies in
hardware versus software) one of the two architectures is chosen. Regardless
of architecture, a backplane bus is involved to connect all internal
elements of a compute node to allow communications and control by a CPU. The
most popular and widely used backplane bus architecture in standards-based
equipment is the Peripheral Component Interconnect (PCI). With CompactPCI,
developers are given the best of both worlds with the commodity availability
and pricing of PCI silicon and a ruggedized form factor.
The PICMG 2.16 architecture is changing the way systems are being
designed by overlaying an Ethernet-based, packet switching architecture on
top of CompactPCI to create an Embedded System Area Network (ESAN). Defined
as a dual redundant star architecture, the PICMG 2.16 standardï¿½s use of
Ethernet provides node-to-node connectivity through redundant Gigabit
Ethernet paths managed by high-performance Ethernet switch fabrics. In
short, the chassis-based network resembles a typical Ethernet network found
in most networking environments.
As demands on performance metrics, such as density, throughput, cost, and
reliability continue to increase, moving system traffic from a shared-bus
architecture such as PCI to an embedded Ethernet backplane such as PICMG
2.16 becomes necessary. Deployments of PICMG 2.16 systems are underway and
can offer extended capabilities and performance an order of magnitude above
current standards-based architectures.
PICMG 2.16 is ideal for most any application now based on the popular
CompactPCI form factor but seeking much higher reliability, performance, and
scalability. In fact, compliant first-generation switch fabrics already can
deliver performance increases up to 40 times that of current PCI
architectures ï¿½ with more reliability and higher chassis density. And with
an unprecedented number of vendors having collaborated on standardization of
the spec and compliant products already filling the pipeline, there is a
wide array of 2.16 products on which equipment manufacturers can build their
PICMG 2.16 FOR THE SERVICE PROVIDER
Faster throughput rates and higher scalability may be enough to sell the
advantages of 2.16 to the equipment developer, but what about the service
provider? What does this ï¿½backplaneï¿½ architecture mean to those looking
to enter new markets and/or enhance or expand existing networks, all while
adhering to their companyï¿½s return-on-investment model? And why is Gigabit
Ethernet inside network equipment so important?
The PSTN is migrating to a packet-switched infrastructure. Overall,
aspects of the telecommunications network are moving towards an Ethernet/IP
model, including 10G Ethernet being used in the metro network and as a
carrier technology, WI-FI networks being developed with the 802.11
standards, SS7 networks offloaded to shared IP networks, and storage
Amidst this move towards Ethernet/IP, each piece of network equipment
needs to keep relative pace with other segments to ensure the reduction of
traffic bottlenecks. Since equipment based on PICMG 2.16 can perform at
levels 40 times that of existing architectures, it is more capable of
keeping pace in the overall network. And with the ongoing growth of
Ethernet/IP-based networks and equipment, there exists the capability of
providing a common networking scheme from end-to-end for any connection,
whether itï¿½s data, voice, or signaling. This commonality is where true
performance and network interoperability advantages will take place.
While the chassis performance metrics for PICMG 2.16 equipment are
attractive, operational advantages for the service provider can be equally
attractive. For example, PICMG 2.16 allows for significantly higher slot
density or capabilities capacity in a typical rack footprint. Higher slot
density allows for more functionality to reside in the equipment and, in
turn, a higher call volume in the same fixed physical space. Higher
provisions for density translate into a greater return-on-investment on
leased central office space, allowing the service provider to control costs
by realizing a more efficient, more scalable call volume and a better
management of lease expansions.
PICMG 2.16 systems also provide increased reparability on a per slot
basis, without affecting other elements in the chassis. Similar to an
enterprise network, it is not necessary to shut down the entire network to
fix one element of the network. This all-important feature reduces mean time
to repair (MTTR) specifications for network equipment through hot swappable
blades and enables a higher degree of uptime.
Since PICMG 2.16 is a standards-based architecture, many of the silicon
elements within the system are produced in higher volumes than those found
in proprietary architectures. Therefore, the risks associated with single
sourcing of equipment, or component obsolescence, are reduced dramatically.
Finally, because of the large number of equipment vendors who
participated in the standardization of 2.16, less than nine months after its
ratification over 30 vendors are already offering 2.16-compatible products.
Many equipment manufacturers worldwide have either adopted 2.16 or are
looking to adopt it in the near future. With this overwhelming acceptance by
equipment vendors, service providers are beginning to find a wide variety of
competitively priced, readily available 2.16-based systems to choose from.
THE FUTURE OF ETHERNET AND 2.16
PICMG 2.16 provides a viable Gigabit speed backplane architecture for
equipment manufacturers. Using this architecture, service providers are
offered true next-generation systems for 2.5G and 3G wireless infrastructure
equipment, multimedia gateways and softswitch elements, signaling
infrastructure equipment, call processing servers, integrated access
devices, and more. Each application shares performance demand for
capabilities such as high availability, density, reliability, and
throughput, all which PICMG 2.16-based equipment is designed to address.
For the future, Ethernet in the backplane is already evolving from
Gigabit speeds with a next-generation standard called PICMG 3.X, which is
being defined with both Gigabit and 10G Ethernet speeds in mind. As Gigabit
Ethernet proliferates as the common enterprise LAN connection, 10G Ethernet
begins standard deployment as a carrier technology, and Ethernet/IP
continues as the common communications mechanism, we will begin to realize
overall network commonality with performance and deployment benefits coming
Ed Bizari is director of marketing at Performance Technologies, Inc.
PTI is a leading supplier of packet-based telecommunications and networking
products for wireline, wireless, and next-generation Internet Protocol
networks. The company provides carrier-grade technology solutions, including
network access products, Ethernet switching products, and integrated SS7/IP
solutions to network operators, equipment manufacturers and systems
integrators worldwide. For more information visit the Web site at www.pt.com.
To The August 2002 Table Of Contents ]