The PICMG consortium of over 450 companies, published the AdvancedMC base specification,
AMC.0, in January of this year, and AMC modules and chassis are here today.
A subset of PICMG’s AdvancedTCA specification for telecommunication and
data center computing hardware, AdvancedMC modules are being adopted in platforms
ranging from AdvancedTCA platforms to custom designs, blade servers and beyond.
They are being considered by companies in the military and commercial markets
as well as those in the enterprise and telecommunications space. AdvancedMC
modules bring a truly modular approach to systems design that no other mezzanine
form-factor, including PMC, has offered in the past.
AdvancedMC (AMC) modules are versatile enough to range beyond the AdvancedTCA
(ATCA) architecture. Although AMC is still early in its market evolution and initial
market data indicates that AMC sales are a small percentage of overall ATCA sales,
it is becoming evident that the opportunities for deploying AMC modules extends
far beyond its application in ATCA systems.
Not all Telecomm Equipment Manufacturers (TEMs) are embracing ATCA wholeheartedly.
There are some TEMs that are not moving their server systems aggressively toward
the ATCA architecture, for the obvious reason of their large investment in their
own proprietary telecommunications platforms. AMC modules, however, give those
TEMs the advantages of a telecommunications open-standard architecture while allowing
them to retain their proprietary base architecture.
AMCs—Fit for Telecom
The AMC’s form-factor has been designed to be as flexible as possible for
a wide range of telecommunication computing and I/O applications in single-wide,
double-wide, half-height and full-height configurations. One advantage of the
AMC form-factor is its larger surface area—about 14 percent larger than
the PMC form-factor—resulting in more room for components. The AMC specification
supports up to 60W of power in all configurations so processor AdvancedMCs aren’t
starved for power.
As telecommunication networks move toward voice-over-IP (VoIP) and 10 Gigabit
Ethernet, the older parallel or PCI-based products such as PMCs fall behind. The
serial fabric interconnect specifications for AMC support up to 12.5 Gbits/s,
and AMC can support a range of protocols including PCI Express and Advanced Switching,
Gigabit Ethernet, Serial Rapid I/O and InfiniBand.
The built-in hot swap capability of AMC modules seems tailor-made for the telecommunications
industry. Unlike its PMC predecessor, the AMC module is front-loadable, sliding
into a chassis, rather than being installed piggyback style in the internals of
a carrier. An AMC module is field-replaceable and upgradeable and does not require
the entire system or blade to be removed. In this fashion, an ATCA or proprietary
blade consisting of multiple AdvancedMCs is flexible, scalable and highly available.
Other important features for telecommunication applications are AMC’s support
for the Intelligent Platform Management Interface (IPMI) as standard and Electronic
Keying (E-Keying). These features allow dynamic configuration of AdvancedMCs as
well as giving them a monitoring and alerting subsystem, both important for ensuring
maximum equipment uptime.
AMCs Open the Way
Current telecommunications opportunities, like VoIP, are pushing the industry
to adopt an open-standard modular computing architecture to build scalable platforms.
The TEMs are starting to move away from pure hardware plays and are looking more
like software companies as they concentrate more of their corporate resources
into their core competencies, like soft switches for example, and leaving the
building of telecommunication hardware to hardware companies.
So where do AMC modules fit into this picture? The open-standard modular architecture
of AMC modules seem almost ready-made for companies that want to leverage their
proprietary server architectures into a telecommunications market that is increasingly
looking for open-standard equipment.
An open-standard modular computing architecture like AMC is based on the idea
that multiple equipment suppliers will be able to compete with one another to
supply the market with inventive products at the best possible price. Ideally,
telecommunication system companies building modular systems want to have multiple
suppliers for each component in their systems—second and third sourcing
as much as possible—so they can take advantage of market efficiencies.
The critical supply chain issues of sufficient inventories, lead-times, quality
and yearly cost reduction commitments are becoming more prominent factors in the
buying decisions TEMs are making. The whole telecommunication manufacturing environment
is starting to look more like the IT industry than the traditional embedded industry.
Indeed, companies that supply major IT systems are very good with supply chain
issues––companies like HP, IBM and Sun. All three manufacturers have
announced strategies to compete effectively in the telecommunications market.
Of course the TEMs have recognized the advantages of open-standard modular computing
and they are adapting their proprietary platforms to support AMC modules. AMC
modules are a natural fit for the proprietary blade servers that the large computer
manufactures are marketing today. This adaptation essentially gives them the best
of both worlds—leveraging a rich ecosystem of open-standard modular AdvancedMCs
and extending the life of their investment in proprietary equipment.
A Bit about Blades
Blade servers and carriers fit into a blade chassis, with the slim, hot swappable
blades acting as separate servers with their own processor, system memory, hard
drive, network controllers, operating system and applications. A blade server
simply slides into a bay in the chassis and plugs into a mid- or backplane, sharing
a common power supply, fan, floppy drives, switches and ports with other blade
All the critical components of a blade server, like cooling systems, power supplies,
Ethernet controllers and switches, mid- and backplanes, hard disk drives and service
processors can be made redundant or hot-swappable. Removing a server for maintenance
simply means sliding the blade out of the chassis.
In advanced blade server systems, the software side of their operation is enhanced
as well. Slide a blade into a profiled bay—the system automatically loads
a designated operating system and application image into the blade; the server
is designed to get up and running with no human intervention. Under software control,
a spare blade can easily replace a failing blade or help handle peak loads. These
uptime and redundant features are just what the telecommunications industry relies
on every day.
Advanced blade server systems offer smart ways of achieving highly sensitive maintenance.
Some blade-server components can alert a systems management processor of impending
failure hours or even days before failure occurs. Advanced diagnostics point a
field technician directly to a failing part for quick, efficient replacement.
Adding a new server generally involves nothing more than sliding a new blade into
an open bay in the chassis. Each blade in a chassis is really a self-contained
server, running its own operating system and software. Sophisticated cooling and
power technologies can therefore support a mix of blades, with varying speeds
and types of processors.
AMCs Meet Blade Servers
As advanced as blade servers are, they are still based on proprietary hardware.
Their flexibility allow them, though, to accept carrier cards that can be configured
with just about any open-standard architecture allowing end users to tailor their
servers just about any way they need.
For example, SBS Technologies and IBM Engineering and Technology Services
are developing an AMC carrier blade for IBM’s BladeCenter family. This
carrier board will extend the BladeCenter server’s ability to support
industry standard, robust AMC-based processors, I/O and WAN monitoring capabilities
for transport plane-intensive applications. These include wireless and signaling
gateways and the breadth of network interfaces used in the telecommunication
industry (Figure 1).
IBM’s BladeCenter carrier blade for AMC modules offers a great deal
of flexibility for system designers. For example, take a BladeCenter carrier
card and populate it with a processor AMC module (Figure 2), add an AMC video
display module and an AMC Ethernet network module, or other I/O, and you have
the makings of a powerful, open-standards-based AMC telecommunications server
blade. Other blade server vendors like HP and Sun won’t be far behind
in using this approach for opening up their proprietary systems either.
If you consider the architecture of the AMC standard—hot swappable, with
serial interconnect and manageable—it lends itself nicely to a stand-alone
system. This is exactly what many companies are working toward.
MicroTCA Releases AdvancedMC’s Potential
PICMG’s MicroTCA system architecture specification consists of AMC modules,
all connected together by a common backplane, sitting in a MicroTCA chassis. AMC
micro-blades can provide I/O and processing power to a stand-alone MicroTCA system
without needing a carrier card, just as ATCA and proprietary server blades provide
those same features to an ATCA system.
The MicroTCA specification has the potential of addressing not only telecommunication
applications, but also applications not particularly suited for ATCA, but that
require a similar level of system performance and features. For example, while
the ATCA architecture is aimed at core telecommunication network operations, the
MicroTCA architecture will be better suited for lower-cost network edge and enterprise
The first MicroTCA demo systems were on display at SuperComm in June of this year
and the final specification is projected to be ratified by the PICMG committee
sometime in early 2006.
The PICMG MicroTCA committee is still working through the details of the MicroTCA
specification and their goal is to build a platform specification that meets the
cost, size and modularity needs of applications not currently addressed by the
ATCA specification. These applications could vary widely, including telecommunications
network applications, enterprise network applications or even consumer applications.
Among the many possible types of chassis configurations, the MicroTCA committee
is considering at least two. One is a possible 4U 19-inch rackmount chassis, which
can support up to 10 full-height AMC modules and could also support a mixture
of the other form-factor AMCs. The other possibility is an 8-inch cube chassis
configuration that could hold up to 12 half-height AMCs.
The MicroTCA specification, like the ATCA specification, will likely support a
range of possible protocols across a serial backplane. The first systems on display
at SuperComm 2005 supported a Gigabit Ethernet backplane and future systems are
projected to support PCI Express and Advanced Switching, Serial Rapid I/O and
10 Gigabit Ethernet.
Probably present in most MicroTCA configurations will be an AMC Virtual Carrier
Manager (VCM) responsible for powering, managing and connecting up to 12 AMC cards
in the MicroTCA chassis. The VCM will provide IPMI system management and a fabric
switch with up to 60 lanes of connectivity with various configurations supporting
Ethernet, PCI Express and other protocols.
AMC modules have the potential to extend the usefulness of proprietary telecommunication
systems, and at the same time, they are not limited to solely serving as adjuncts
to the ATCA architecture. When the MicroTCA specification is finalized, AMC-based
servers will have the potential to branch out beyond the telecommunication market
into the data center, military and other markets.