July 1999
Industrial Computers: From Auto-Assembly Lines To Phone
And Internet Lines
BY RON GROEN
TURNING POINT OF COMPUTER TECHNOLOGY
Ten years ago, CTI applications were non-PC-based. Most network service providers had
their own versions of proprietary equipment, running applications on closed computer
systems. With these user unfriendly systems, the process of integrating computer and
telephone systems with data networks was often difficult, expensive, time-consuming, and
required diversified skills and tools.
But today, closed systems used in telecom (and more recently Internet) OEM applications
are starting to go the way of the rotary phone and Apple II computer. Proprietary systems,
which were once leading-edge technologies, are slowly being replaced by open computing
platforms. According to Natural MicroSystems,
high-availability technologies specifically CompactPCI and hot swap are
enabling the migration of telecommunications infrastructure equipment from proprietary
hardware and software architectures to mass-market PC hardware and industry-standard
software operating systems like Windows NT.
This trend is most evident in the Open Source for Open Telecom initiative recently
endorsed by the PCI Industrial Computer Manufacturers Group (PICMG), a consortium of more
than 450 industrial computer product vendors. This group collaboratively develops
specifications for PCI-based systems and boards for use in industrial and
telecommunications computing applications. The initiative they endorse is aimed at
providing the public, in source code form, the hot-swap software infrastructure for
CompactPCI and the software for telecom circuit switching and resource management using
the CT Bus (H.100/H.110), specified in PICMGs CompactPCI Computer Telephony
Specification.
PICMG is paving the way for industry computers
used in datacom equipment. Previously used in monitoring systems for utilities and
industrial plants, semiconductor thermal processing machines, seismic data acquisition,
processing and interpretation systems, and much more, the industrial computer is becoming
the ideal choice for Internet/telecom OEM equipment applications for several reasons.
Ruggedized Hardware
From a design perspective, industrial computers are an ideal fit (no pun intended) for the
datacom OEM market. Most of the design features found in industrial chassis, backplanes,
and single board computers (SBCs) match the needs of todays OEM applications. For
example, the industrial computer chassis rackmount design fits into a standard 19"
cabinet used within the datacom market. Other chassis are available in benchmount, tower,
and rackmount configurations. Industrial computer chassis have many physical attributes,
including a metal front panel; ruggedized design (which can withstand adverse heat, shock,
and vibrations); redundant power supplies; cooling; load sharing;
hot-swappable/hot-pluggable, compliant, passive backplanes, fans, power supplies, and
media; and front-accessible peripherals all of which are needed in datacom
equipment.
Features found in industrial SBCs which mirror Internet/telecom equipment requirements
include front or rear I/O access; PCI/ISA BUS-compatible; dual, universal serial bus
(USB); Intel Celeron processor speeds of 300, 366, or 400 MHz; system management
functionality compliant with IPMI (Intelligent Platform Management Interface); and more.
Product longevity and durability are other benefits for industrial PCs used in datacom
applications unlike other industries, where the PC technology is often obsolete in
a couple of years after being designed. The industrial computer market offers products
designed for the long haul with an average life span of up to five years. What this means
is Internet and telecom OEMs can rest assured that the industrial computer products that
they are utilizing today will be available in the future. How many people do you know who
bought a PC a few years ago that is now nothing more than glorified paperweight?
In addition, mass production of a product cuts down on quality. Industrial computers
are often built with more durable and often more expensive components with
higher mean time between failure (MTBF) levels than computers used in the commercial
market. What this means is greater uptime for the end user.
Bandwidth
As Moores law which states that the pace of technology change is such that
the amount of data storage that a microchip can hold doubles every year, or at least every
18 months became a reality in the computer market, so did the demand for higher
bandwidth applications. These include diagnostic imaging equipment that processed
conventional and digital X-ray, computed tomography, magnetic resonance, and ultrasound
applications. The need for high-bandwidth applications has also carried over to the
equipment for high-speed Internet, intranet, extranet, e-commerce, and streaming (audio or
video) applications through high-speed access technologies including asymmetric digital
subscriber line (ADSL), hybrid fiber coaxial (HFC), wireless local loop (WLL), and
satellite.
BUY, DONT BUILD
Consideration of pricing for development and design of proprietary systems is another
factor, which has led to the growth of the industrial computer in the datacom market. Why
build a proprietary system, when you can buy an open industrial system (from a growing
list of manufacturers) for less money? If companies like Lucent
, Alcatel , and Nortel Networks could build proprietary computer
systems for the same price they pay for industrial PCs, they would do so. With lowered
development costs attributed to open systems, OEMs can design software around computers
and NOT design computers around the software. What this means is OEMs can
standardize computers and reallocate their engineering resources to new software
developments, thus focusing on their core competencies.
Another benefit the industrial computer provides the Internet/telecom market is the
continuous availability (up to 99.999 percent of the time) operation of the equipment.
This differs for fault-tolerant, proprietary systems, which use redundant hardware and
software to provide non-stop operations. By maximizing the availability of systems and
applications through decreased downtime during scheduled maintenance work and unexpected
system failures, industrial computers can just about guarantee an Internet server, switch,
or router will be operational.
SYSTEM MANAGEMENT
One final driving force of the industrial computer is its system management
interface, found in the intelligent platform management interface (IPMI). Using the I2C
BUS, IPMI is the common interface of intelligent hardware that is used to monitor several
physical characteristics of the hardware (in this case the Internet server, platform,
etc.), including temperature, voltage, fans, power supplies, and chassis. These monitoring
abilities provide information that enables system management, recovery, and asset tracking
that help drive down the total cost of ownership (costs outside of the initial software
and hardware of maintaining a system) for network service providers and thus,
reducing downtime.
According to Intel Corp., IPMI specifications
give IT managers access to platform management information and control features that allow
more accurate prediction of hardware failures, diagnoses of hardware problems, and
initiation of recovery actions. IPMI allows IXCs, ISPs, ITSPs, etc., to determine
the health of their equipment, whether it is running normally or it is in a
non-operational mode. Equipment based on IPMI uses intelligent or autonomous hardware that
remains operational even when the processor is down, so the platform management
information is always available to the IT manager.
IPMI interfaces enable telecom equipment to be accessed not only by management
software, but also by third-party, emergency management add-in cards, and even other
IPMI-enabled servers. This allows network service providers to gain flexible and
interoperable access to vital platform management information. System-to-system monitoring
or management via a connected server is becoming increasingly important as IT managers
deploy complex system topologies such as clusters and rackmount configurations. In
addition, the scalable nature of IPMI enables the architecture to be deployed across a
server product line, from entry to high-end servers, and gives IT managers a consistent
base of platform management functionality upon which to effectively manage their
equipment.
According to industry sources, the telecommunications equipment market is a
$150-billion market, with approximately 90 percent of the business built on proprietary
hardware and software. This means there is a tremendous growth opportunity for companies
willing to enter and succeed in this market. Ten years ago, who would have thought the
industrial computer found in the automotive plants in Detroit would be connected so
closely to the computers found in todays Internet/telecom equipment? The
industrial computer once at home on the auto-assembly line floor in manufacturing
facilities in Detroit, is now finding a new home within the Internet/telecom (datacom)
market. Whether it is being used as original equipment manufacturer (OEM) components in
servers, switches, routers, or voice/data switches, todays datacom OEMs are
demanding the ruggedized hardware, price, bandwidth, availability, and system management
previously only available in industrial computers.
Ron Groen is worldwide vice president, sales and marketing, for Texas Micro, Inc.
Texas Micro is an ISO 9001-certified manufacturer specializing in the design, production,
and integration of open system computing platforms for telecom, Internet, and
industrial-specific applications, including imaging, data acquisition, industrial control,
and network applications. For more information, visit the companys Web site at www.texasmicro.com
|
Compatibility
Testing Protects Customers And Suppliers BY DON HOWELL
According to many experts, the Internet telephony market is expected to grow anywhere
from 100 to 150 percent annually for the next five years, yielding tremendous market
opportunities for VARs, integrators, and suppliers. Multimedia desktop PCs and
high-availability industrial-strength models, either ISA or PCI running Windows NT or
UNIX, will continue as hardware platforms of choice for the next several years. However,
explosive market growth, combined with a lack of established standards, has created a risk
environment for both suppliers and users of industrial computers and peripherals.
Understandably, the high cost of sales, low margins, faster time-to-market and other
competitive pressures make it very unattractive, if not prohibitive, for industrial
single-board computer suppliers to physically verify operation of their SBCs with all of
the hardware peripherals and software for which they claim compatibility. Suppliers and
users alike falsely assume that, because an SBC works with some popular programs and
peripherals, it must work with them all.
Users can raise the odds of interoperability by buying all of their system elements
from the same source however, this is not always possible. Another way to ensure
compatibility is to require that the supplier provide test reports, performance
benchmarks, or other empirical data to support any such compatibility claims. For the
supplier, compatibility verification testing may be accomplished either through an
in-house self-certification program, or by outsourcing the job to an independent test lab.
INTEROPERABILITY TESTING
Lets first consider in-house programs. The self-certification process begins at the
breadboard/prototype stage, and continues through internal beta sites that effectively
isolate most basic operational compatibility problems among processors, chipsets, BIOS,
peripheral controllers, operating systems, and more.
Standardized SBC diagnostic tests of COM ports, LPT ports, onboard controllers, IDE, and
floppy controllers are used to verify that everything is performing correctly. Read and
write tests verify that various hard drives work with all the various operating systems,
from DOS and Windows NT to SCO UNIX.
Despite such seemingly thorough procedures, sometimes a deeper level of investigation
is needed to isolate the root cause(s) of an embedded compatibility problem. Eighty to 90
percent of SBC compatibility problems are usually traceable to the BIOS, and are then
corrected by the original supplier, and reinstalled.
Once an SBC supplier reaches a certain sales plateau, the economies tip in favor of
performing much or all of the compatibility testing program in-house, which allows greater
control over the types of hardware and software that can be tested. It also allows quicker
response to customers, both large and small. But for suppliers who simply cannot support
the resources needed to develop, operate, and maintain a viable in-house compatibility
certification program, subcontracting the job to an independent testing lab is another way
to get similar results.
Fewer than half a dozen independent testing laboratories specialize in this area, and for
a supplier, the price of admission can be quite high often $50,000 or more to
certify a single SBC series. One such independent testing laboratory based in Los Angeles,
CA offers a complete test regimen that takes the interactive hardware and software
compatibility issue as far down the road as the client wishes to go.
The product certification process ends with the award of a silver, gold, or platinum
seal representing good, better, or best ratings for demonstrated compatibility with
currently available hardware and software products within the client companys target
markets. For example, the mid-range rating might certify compatibility of the tested
product with at least 80 percent of the available products with which it could be
conceivably used. Though expensive, what makes this proposition attractive to the supplier
is the probability of increased sales that will result from certified compatibility with
literally hundreds of commonly available hardware and software products.
The various test phases of a mid-range-level compatibility certification, for example,
include a large representative sample of currently available hardware and software
products, laid out in a very structured, organized testing environment, such as:
Phase 1: Operating System Software Testing.
Phase 2: Hardware Peripherals Testing.
Phase 3: Hardware Interoperability Testing.
Phase 4: Software Testing.
Phase 5: Network Testing.
Phase 6: Performance Benchmark Testing.
CONCLUSION
The old adage that you get what you pay for certainly applies to the industrial SBC
market. The annoyance of pulling a board out of the box, installing all the hardware and
software, and then having it not start up, is not unusual with untested boards. For the
supplier, contracting an independent testing laboratory, or doing all of the testing
in-house, may seem too large an investment for a fledgling company. In
reality, it is a fraction of the cost of taking a high-performance SBC to market in the
first place. In the final analysis, how does the supplier or user benefit if the product
doesnt work?
From the users standpoint, the compatibility problem is slowly improving as more
SBC manufacturers build products in compliance with industry standards, such as those
defined by the 350-member PCI Industrial Computer Manufacturers Group (PICMG). In the
absence of more objective data, users are encouraged to look for PICMG-compliance as a
good indication of probable compatibility among industrial PC hardware, software, and
peripherals.
Don Howell is a compatibility testing supervisor for Industrial Computer Source.
Founded in 1985, Industrial Computer Source manufactures and stocks PC platform solutions
supporting Intel processors and compatible operating systems. The company specializes in
rackmount computers, chassis and enclosures, single board computers, and custom and
ruggedized computers and chassis. Products are targeted for IP telephony,
telecommunications, voice processing, broadcasting/convergence, medical, and industrial
automation industries. For more information, visit the companys Web site at www.indcompsrc.com |