If you've been listening to your IS staff gossip around the cappuccino machine, you've heard them boast that their departmental LAN is running on a near-perfect operating system called Linux. Pay attention; in three years you may be installing it on all your company's servers.
Why the wait? Although this Unix look-alike has gained an outspoken, loyal following among the technical intelligentsia, it's not yet the mission-critical server platform. But the features prized by CIOs are coming, says William J.
Peterson, research director for client and embedded operating environments at analyst firm International Data Corp. in Framingham, Massachusetts. (a sister company to CIO). He says that Linux will move into the mainstream in three years, predicting, "The question will become, 'Are you considering Linux?' instead of, 'Why are you using Linux?'"Some CIOs have already considered Linux and decided not to wait. These early adopters have discovered the worst-kept secret on the Internet: a free operating system that rarely crashes, runs on hardware ranging from a 386-based PC to a Sparc-based Unix server, is easy to debug and modify, speaks IP fluently, looks like Unix in terms of how it's administered and keeps getting better and better thanks to a grass-roots development effort involving thousands of programmers worldwide. It was fathered by Linus Torvalds, a Finnish programmer who, when he was a college student in 1991, wrote the operating system that was the seed for Linux (he named it after himself). He subsequently coordinated a loose confederation of global volunteers who expanded Linux, adding capabilities like SCSI drivers and compliance to the U.S. government's Posix standard that boosted the operating system's usefulness.
In the method of its birth, Linux is a cousin to well-known Internet tools such as the Apache Web server. These freely distributed programs are called open-source software because not only are they available in their compiled versions, but you can also download the source code and customise it. Thousands of developers have done so with Linux and then posted their improvements on the Internet, resulting in everything from a multimedia-ready distribution to a symmetric multiprocessing version. Current development efforts work on adding capabilities ranging from a Windows NT file system to asynchronous transfer mode (ATM) networking.
Other developers have decided that the way to improve Linux is to focus on support. Several companies, notably Red Hat Software Inc. in Research Triangle Park, N.C., and Caldera Systems Inc. in Orem, Utah, sell so-called commercial versions of Linux, which boil down to a CD of the free operating system bundled with other open-source programs like sendmail, proprietary utilities, documentation and technical support, all for as little as $50. It's a deal that more than 250,000 people buy each year.
But that number masks the true popularity of Linux, since it's impossible to track the free versions. Red Hat published an estimate earlier this year that pegged the number of users at 7.5 million. IDC's Peterson estimates that the number could range as high as 10 million seats, calculating that annual growth has risen from 20 percent to 50 percent in the last year.
For CIOs, buying a commercial version is the way to go because it provides one thing that downloading Linux off an FTP site does not: accountability. While Peterson acknowledges that the Internet-community support model works, he also points out that CIOs want to have "someone to throttle."Vendors standing behind Linux is one of four factors that need to be in place before Linux becomes a mission-critical server, according to Peterson. That's happening quickly. For example, Caldera and Red Hat are creating training programs, a sign to CIOs that the support model is changing into a full-fledged services program. They're also building add-ons that address CIOs' concerns, such as easier administration and integration into an existing infrastructure.
The second factor is progressing more slowly: support from a hardware-platform owner, such as Sun Microsystems Inc., Hewlett-Packard Co. or IBM Corp. Although there are many public projects that have or will port Linux to processors such as the PowerPC, Alpha and MIPS, no major system vendor has stated that it will offer Linux to its customers as an alternative operating system. But there are signs of interest in this fast-growing OS. Sun recently joined Linux International, a nonprofit organisation promoting Linux through publicity, trade-show exhibits and grants to programmers upgrading the software.
The outlook for Peterson's third factor is mixed. Before Linux becomes a trusted part of a CIO's toolbox, it has to be supported by the application vendors most important to CIOs. The good news is that more and more core business applications will be ported to Linux; in the near future, you will be able to run Oracle Corp.'s Oracle 8, Netscape Communications Corp.'s Messaging Server, Netscape Directory Server and Informix Corp.'s Informix-SE on Linux.
What's driving the migration is pure numbers. According to Greg McCullough, spokesman for Control Data Systems Inc. in Arden Hills, Minnesota, approximately 80 percent of large corporations have Linux running somewhere.
That figure convinced Control Data to port its IntraStore Server 98 messaging software on the platform.
But for every big name entering the Linux playing field, there is another big name watching from the sidelines. Take SAP AG, for example. Its representative says the company has no immediate plans to create a Linux version of its enterprise resource planning software. The reason given: platform decisions are based on customer demand.
That statement raises the spectre of the chicken or the egg. As a CIO, do you wait to switch to Linux until your third-party applications run on it, or do you commit to Linux and start asking your suppliers for Linux versions? If you chose the latter course, you don't have to waste resources on an idle trial system. If you consider Internet access or your corporate e-mail to be mission critical-and who doesn't-then Linux can play a significant role in your infrastructure right now.
Mark Lehrer, MIS director at Health Design Plus Inc., a medium-size health-care management and claims-administration company in Hudson, Ohio, chose to stay ahead of the curve. He installed Linux in all three of his company's offices, relying on it for routing duty and file, e-mail and proxy services. He runs Linux because "simply put, it gets the job done," he says. "It fits in well with our other Unix system and makes Windows 95 clients happy." That other Unix system, an HP-UX machine, exists because Health Design Plus needs a database.
Eventually, the market caught up to Lehrer and Linux. "Now that industry-standard databases are available," he says, "I will be trying them out."Kenneth R. Jorissen, information and control-systems manager at Nonvolatile Electronics Inc., an R&D company working with giant magnetic sensing materials in Eden Prairie, Minnesota, is in a similar situation. He replaced an unreliable Windows NT server with the Linux system that now supplies all his users with services such as file, print, Internet and Dynamic Host Configuration Protocol (DHCP, a way of assigning IP addresses on a first-come, first-served basis). What's holding him back from of a pure Linux installation is that a key laboratory test-and-control package doesn't run on Linux.
Achieving the fourth and final factor may prove to be the most difficult. The proponents of Linux "need to lose the hacker OS mentality," says Peterson.
"They need to make their presentation acceptable to Dilbert's boss." Some have tried and met with great resistance; other efforts have been more successful.
Groups like Linux International, which take a helping-hand approach to shepherding the Linux community, have been generally accepted. But an attempt by two little-known software companies to create a standards body generated a firestorm of protest. Its aim was to channel Linux development along certain guidelines thought to appeal to business customers. Linux fans denounced it as a sure-fire way to destroy the open-source development that has caused Linux's rapid innovation.
For users like Health Design Plus's Lehrer, that open-source nature is Linux's biggest competitive advantage. "Its overwhelming strong point is its user community-without any doubt," he says. "I have several instances where I had problems with a device and was able to exchange e-mail with the author of the device driver. This is unheard of from Microsoft or Santa Cruz Operation Inc."But looking at it from the outside, open-source development can seem chaotic.
"All the Linux distributions have problems with backward compatibility. That's the nature of hackers and hobbyists messing with the kernel," explains IDC's Peterson. "They don't document it or test it. Someone will take [their work] and incorporate it into a Linux distribution. You can't manage that [kind of development] efficiently."In all likelihood, the task of polishing Linux into a professional product will fall upon the commercial companies that have sprung up. They control the contents of the Linux distribution they support, so they can test new features and write manual pages before they ship the CD.
Eventually, the four factors that are holding Linux back will fall into place, Peterson says. If that happens this year, he calculates that 1999 will be spent on test projects and the following year on implementation, making 2001 the time frame for Linux's acceptance in the enterprise.
(Amy Helen Johnson covers technology from the Silicon Forest in Redmond, Washington. She can be reached at firstname.lastname@example.org.)Reducing Manufacturing TimeCan the assembly line be the shortest distance between two points?Production EngineeringGetting a product from the design stage through prototyping and production is a protracted process. Despite the wide availability of computer-aided design (CAD) and computer-aided manufacturing (CAM) tools, it still takes the automotive and aerospace industries five years or more to bring out a car model.
That's because the production engineering phase is still a hands-on paper and pencil operation even in the most sophisticated high-tech manufacturing companies. Engineers must hand-build working prototypes, design the tools, create a manufacturing and assembly line, and then spend months fine-tuning every phase. Wouldn't it be nice to build working prototypes, lay out manufacturing lines on a computer screen, program the robotics devices and actually detail every minute task in a simulated production environment digitally?With the new computer-aided production engineering (CAPE) tools from Tecnomatix Technologies Ltd., an Israel-based software firm, automotive giants like General Motors Corp., Ford Motor Co., BMW AG and Honda Motor Co. Inc., among others, have built "digital factories" and have begun to shrink the time it takes to design and build a new car model from 60 months to 24 months. Changing over production lines can now be done in 48 hours instead of six to eight weeks, saving millions of dollars and delivering better products at lower cost.
"We've spoken with many of the world's largest manufacturing companies, and they say it works," says Bruce Jenkins, vice president of Daratech Inc., a market research and technology assessment firm in Cambridge, Massachusetts.
Once a manufacturer has created a digital library of parts and components (including highly specialised robots, automated tools and fixtures), engineers can drag and drop them into a model of new assembly lines on their computer screens and literally see the product manufactured before their eyes. They can zoom in on any individual cell and time the operation, detail all manual operations and streamline the ergonomics so that they are compliant with government safety regulations.
The digital production environment also saves time because it can be set up concurrently with the design phase, says Jenkins. "There is a lot of communication between the two processes, as well as an overall shortening of the time to market, so that companies don't lose a window of opportunity," he says.
Although Jenkins says that Tecnomatix is the clear leader, Deneb Robotics Inc. in Troy Michigan, acquired by Dessault Systems in Paris, distributed worldwide by IBM Corp., isn't far behind with its own end-to-end digital manufacturing tools suite. With its success on a large scale with planes and automobiles, Jenkins notes, expect to see it filtering down to the manufacturing of smaller applications, including consumer electronics, home appliances and computers.
A Cordless Phone for the Office
Generally the announcement that two companies have made their products work together makes us yawn. But every so often such a pairing creates a startling synergy. Selsius Systems Inc. in Dallas makes PBX systems for IP telephony.
Symbol Technologies Inc. in Holtsville, New York, has been implementing wireless communications systems in the enterprise for years. So put together Symbol's Spectrum24 NetVision wireless phone and Selsius's IP networks, and you have a phone that employees can carry around with them and automatically dial into and out of the company's phone system. The companies are initially targeting retail systems, but any scenario in which employees are rarely at their desks would be applicable.
Selsius's CallManager software runs on Windows NT and integrates with a traditional PBX so that calls travel from one system to another transparently.
The Spectrum 24 phone is about the size of a cell phone and incorporates the International Telecommunications Union's H.323 protocol governing IP teleconferencing. For more information, call +1 800 722-6234 or visit http://www.symbol.com.
Everything Old Is New Again
The problem with y2k remediation is simple: once you're done, all you have is remediated code, not something that works better. To help bring Cobol code forward into this decade, Intercomp Ltd. in Herzliya, Israel, has developed JavaMaker, which provides for the automatic conversion of Cobol applications into Web-enabled Java applications running on multiple tiers or in client/server scenarios (pick your buzzword). The software works with code written in Cobol, PL1 and Natural running on mainframes or any other platform.
In addition to code conversion, it is designed to address the application's interfaces and database structures as well. In its analysis of the system, it retains all features and extensions.
JavaMaker can parse and analyse CICS, ISPF, SQL/DS, DB/2 ISAM, VSAM and IMS databases for deployment in an Internet or intranet environment. Pricing is based on the number of lines of code converted; the personal edition costs US$49,999, which includes a license for 100,000 lines of code. For more information, call +1 714 436-4201 or visit http://www.cobol2java.com.
Getting ERP applications to talk to your other applications can be a time-consuming project, and so-called enterprise integration application tools are sprouting like mushrooms. Their goal is to allow faster setup of interaction between applications and thus nimbler reaction to business and competitive conditions. Visual Edge Software Ltd. in Montreal, has announced Madrid, an EIA tool designed to link SAP R/3 implementations with other enterprise systems, including information analysis, reporting and job-scheduling tools. Madrid also allows tools using industry-standard communication capabilities (Corba, DCOM, Java, Enterprise Java Beans or the Tuxedo transaction processor) to execute R/3 business rules. Madrid runs on Windows NT and starts at $60,000. For more information, call +1 800 609-3301 or visit http://www.visualedge.com.
Can the Spam
If you stored all the unsolicited e-mail messages your employees get over the course of a year, you could fill up a really big hard drive that ideally you could then remove and drop on the foot of the bozo(s) who send them. Computer Mail Services Inc. in Southfield, Michigan, has tackled the problem-at least for those who use Microsoft Corp.'s Exchange-by developing Praetor, a utility that filters unsolicited e-mail, a.k.a. spam. Running on a Windows NT server (not necessarily a separate one), Praetor looks at e-mail at both the protocol and message levels. You can create lists of approved or unapproved text to isolate unwanted mail. Praetor also incorporates a patented technology to identify spam based on the techniques used by those who send out junk e-mail.
As with most plug-in utilities, the software integrates with Exchange and appears as a resource within Exchange's administrative panel.
Praetor (the name is Latin for leader) starts at $295 for networks of 25 users or less. For more information, call +1 248 352-6700 or visit http://www.cmsconnect.com.
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.