Linux, open movement to transform e-business, Zeitler says

January 30, 2002

Bill Zeitler, senior vice-president and group executive of IBM Server Group, spoke at LinuxWorld, saying the innovation of Linux and the open movement will transform e-business. Below are his prepared remarks.

It is a real privilege for me to be with you here today, and I thank you for the opportunity. IBM has had a strong relationship with LinuxWorld and the Linux community for the last several years. I am very grateful to be here at what I feel is a critically important time.

This is a view that is shared broadly across IBM. It comes from the decades we have spent in the information technology industry, and it comes as a result of one of the most comprehensive strategic reviews we have ever undertaken in our company.

When you look at what has gone before, and where we stand today, one thing is absolutely clear -- the technologies of the open movement, and the people of the open movement -- are changing the way the information technology industry operates. By that I mean where and how innovation is done, how companies will be built, and how technology users will find competitive advantage. Those people who deny this, those people who don’t understand this, are on the wrong side of history.

Why do I feel so sure about that? Take a look back. The first 40 years of the computer industry followed a common pattern - a common business model, if you will. From the mainframe to minicomputers to VAX to PCs, companies in the industry aimed to secure the loyalty of developers and clients to a platform - to a set of protocols and tools and processes.

From one technology to the next, the price point changed but the model didn't -- get people committed to the mainframe operating system, VAX, Solaris -- establish control points over a particular set of technologies and use those control points to channel innovation and drive increasing financial return. For decades, information technology was an industry defined by these control points and the boundaries, or barriers, they presented.

Today, however, the developers and systems of the open movement, like Linux, Apache, the protocols of the Internet like html and http, and the Global Grid Forum, have achieved critical mass. We are leaving the era of the proprietary platform and entering the era of the open platform -- of the Web, of Linux and ultimately, of Grid computing -- where IT companies can't defend the barriers they have prospered behind.

While other open technologies preceded Linux and Apache, Linux and Apache took the open development model to a new level. By opening every step of the development process to contribution, Linus Torvalds took advantage of the world's largest R&D resource, free of charge. Unlike proprietary operating systems, which have traditionally been developed primarily by native English speakers, Linux contributors come from every continent and every time zone.

Clearly, contributions to technologies of the open movement are global. Just as clearly, the impact of the technologies of the open movement is pervasive.

Each month, Netcraft surveys millions of Web sites worldwide to see what operating system and Web server they're running. As of December, January's report is not yet posted, Apache is running on 63% of Web sites worldwide, more than twice as many as Microsoft -- and 30 times as many as Sun iPlanet [at 2.8%]. Linux is running on 30% of those sites, three times Solaris and catching up to Microsoft. Open BSD makes up another 6%.

According to IDC, Linux grew at 24% in 2001, making it the fastest growing server operating system. IDC now estimates that Linux will account for 9% of corporate IT budgets in 2002. By the end of 2002, IDC forecasts, based on preliminary data, show that Linux will grow 37% -- in a market in which Windows NT will grow 6%.

IBM’s commitment to the open community is not new. In 1998 we dropped our proprietary Web server in favor of Apache, which is now the basis of our WebSphere product. At the time we made that decision, our WebSphere marketshare was hovering in the single digits. Since adopting Apache, we are nearly neck and neck for the top position in the industry, with 34% marketshare. [BEA has 36%]

Also in 1998, we open-sourced the code to our Java compiler Jikes. In 1999 we open-sourced Jakarta to provide advanced capabilities for Apache, in addition to donating some new XML technologies. Last year we released JFS, the first enterprise ready journaled file system for Linux from our AIX UNIX portfolio. It’s worth noting that AIX growth has outpaced any other commercial UNIX since we started driving this Linux affinity. On our DeveloperWorks site, you can see the dozens of projects we’re currently contributing to, in all areas of Linux.

In addition to serving as a member of the Free Standards Group and continuing to invest money and people in improving and supporting Linux, we’ve contributed significant resources to the development of XML, UDDI, SOAP and other open protocols. And there are lots of other examples.

As some of you will remember, IBM’s president -- and CEO since yesterday, Sam Palmisano -- was on this stage at LinuxWorld a year ago. Sam talked about four IBM customers who were implementing Linux in various ways. Those customers were weather.com, Shell, Lawson, a retailer in Japan, and Telia, the largest telco in Scandinavia. In the intervening 12 months, all four of them have ramped up their Linux implementations.

So, last year we talked about four customers. We had some others, but we picked four. This year, the pool of IBM customers to choose from has grown to around 2500. They range in size from a single ThinkPad to some of the largest supercomputers in the world doing genomic research. But 2500 customer engagements on Linux in a single year by just our company speak to how pervasively this movement has taken hold in the commercial world in which we operate.

In the petroleum sector, in highly parallel supercomputing environments, Linux supercomputers are rapidly taking the place of proprietary UNIX systems. Last year, Sam Palmisano talked about Shell. But it’s not just Shell. Across the sector, companies like Chevron, CGG in France and Western Geophysical and Conoco are taking advantage of this. With clusters of hundreds and thousands of servers all running Linux, many running the cluster management software we moved from our AIX environment.

The same thing is happening in Life Sciences. Structural Bioinformatics, for instance, uses Linux to model the 3D structures of proteins. Linux has reduced their cost per calculation from $28 to $1. Devgen, a Belgium-based biotech firm, is running Linux for genomic sequencing for leads on new drug compounds. MDS Proteomics is another example, running Linux clusters to do drug development. Accelrys is doing drug research on a cluster of IBM eServer pSeries UNIX systems as well as an xSeries Intel cluster running Linux.

In the telecommunications sector, last year Sam Palmisano talked about the fact that Telia had consolidated 70 Sun Solaris servers onto an IBM eServer zSeries. Telia has gone on to increase their mainframe deployment with an additional system - but the point is, a lot of telcos are doing this. Companies like AT&T, Deutsche Telecom, Brazil Telecom, Sonera Entrum, a subsidiary of the largest telco in Finland, CSK in Japan and Netsiel, the IT provider for Telecom Italia are just a few of them.

It is clear that the telco industry is moving toward Linux not only in the data center, but on the network side as well. IBM has had a focus on NEBS compliance in our UNIX server family for years, and we are driving the same focus into the Intel space.

Last summer, IBM announced a couple of direct current products targeted specifically at the telecommunications market. Cambia Networks is just one example of a number of companies using this technology. Cambia is running a completely open, carrier grade wireless Linux solution on eServer xSeries 1U rack systems.

If you want to take a look, you can see our 2U NEBS-compliant server, to be announced later in the quarter in the solution center. We’ll also be launching a NEBS-compliant version of our x330 1U server in the spring.

Open is, clearly, the direction for this sector. At 11:00 this morning, the Open Source Development Lab held a press conference. The OSDL is an independent, vendor neutral group dedicated to guiding Linux for the enterprise and delivering carrier grade functionality. The OSDL has offered a neutral meeting place for open source developers for over a year and have more than 25 active projects underway. The 22 member companies include IBM, Cisco, Nokia, TurboLinux, Red Hat, SuSE and HP. Today, OSDL announced a new working group focused exclusively on enabling carrier grade Linux for the telecommunications sector.

IBM is completely committed to this, from top to bottom. As a matter of fact, our own vice president of server development, Ross Mauri, happens to be the president of the OSDL.

We’re also seeing tremendous Linux adoption among financial services customers. In a panel later today hosted as part of IBM’s LinuxWorld customer day, Salomon Smith Barney will talk about what they’re doing with Linux. But it’s not just Salomon. Among financial services companies, people are seeing the opportunity to consolidate dozens, and in some cases, hundreds, of UNIX servers onto mainframe systems. Places like Merrill Lynch, SIAC, the information technology provider to the New York Stock Exchange, Banco Mercantil in Venezuela, and Shenzen Bank in China and many others.

On the same panel, a representative from LL Bean will talking about how they’re using Linux. Other retailers using Linux include Tommy Hilfiger, Home Depot and Burlington Coat Factory.

Pixar Animation will be there too -- the company that produced "Toy Story 2" and "Monsters, Inc." -- to talk about the work they’re doing with us on Linux. They’re replacing hundreds of SGI graphics workstations with more powerful IBM IntelliStation systems running Linux. Other animation companies are doing the same thing, like Digital Domain which used Linux to create special effects for the movie Titanic. In the entertainment appliance environment, both TiVo and a version of Sony’s PlayStation2 run Linux.

As you may be aware, IBM runs the technology systems behind a lot of the biggest sporting events, like the US Open golf and tennis championships. The most recent is the Australian Open. Linux servers supported the infrastructure, as well as performed the tournament’s draw -- the first time it hasn’t been done by hand. Going back to the Sydney Olympics in 2000, these major sporting sites all run on Linux.

And then there are Web operator kinds of examples. Last year, we talked about weather.com. This year, we’re proud to announce that E*Trade has decided to migrate a number of middleware servers off of Sun Solaris and onto Linux on the IBM eServer xSeries Intel platform. The new system will provide up to three times the user capacity of the current infrastructure.

Other web operator examples include Google, which runs its search engine on 8000 Linux servers, eBay, Akamai and Netscape. It was recently reported in the news that Amazon saved itself $17 million in a single quarter by switching to Linux and Apache for its retail business. Last month in an SEC filing, Amazon announced its first quarter of profit.

A lot of these businesses are running Linux on IBM platforms, but many others are not. Most of these companies can be found listed on Web sites that track Linux adoption. If you want to see for yourself, go to a site called "Companies Using Linux." You can find it by going to Google and using Linux to do the search.

You hear a lot of people say that Linux isn’t appropriate for all applications, and that’s true. You don’t hear me talking about running giant SAP databases on Linux or supporting ATMs. But in areas where it fits, it offers tremendous competitive advantage.

When IBM started this journey, we had fewer than a dozen Linux servers running our own production IT work internally. Today, we have more than 800, and we’re adding them at close to 100 a month. The cost advantages are simply overwhelming.

Our web site www.ibm.com/Linux runs on two redundant, load-balanced xSeries servers running Apache on Linux.
Our intranet site, benchmarked regularly as one of the best in the industry, depends on Inktomi’s search engine on four SMP xSeries servers running Linux.

We run IBM forums, repositories for news on hundreds of topics accessible worldwide, on Linux virtual partitions on two zSeries mainframes.

In our research organization, more than 80 Linux servers have been put to work on Project BlueGene, as we develop our one million processor system, to do protein folding simulation.

It is clear that implementation of these technologies of the open movement drives two kinds of results. Technology users see clear, competitive advantage in the choices they make, and technology companies, like IBM, see clear, competitive advantage in the offerings we’re able to bring to market. Let me offer a example.

In October of 2000, IBM’s server business announced a new initiative called IBM eServer. eServer involved a lot of changes to our business as well as our product lines, but one of its most crucial elements was a commitment to offering people choice in the way they develop and deploy applications. In support of that, we said we would make Linux available across all of our platforms - our zSeries mainframe products, our iSeries products for middle-market kinds of customers, our pSeries UNIX products and our xSeries Intel products - and we have fulfilled that promise.

Since making Linux available on the zSeries, for instance, our mainframe business has achieved five consecutive quarters of significant growth. In 2001, we grew workload and revenue in the mid-teens. This is the first year we’ve seen growth like this -- in fact, any mainframe revenue growth -- since 1989. In the fourth quarter, for example, almost all year to year workload growth was Linux. Linux brought new work, and even new customers, to the mainframe.

Let me give you a few other examples of impact.

Last May, we announced a program to give developers access their own virtual Linux servers on an eServer zSeries over the Web, free of charge. We called it called the Linux Community Development System, and in the first three days, we had 27,000 inquiries. We had to scale up the program to accommodate the demand. We are currently serving developers on virtual servers in 75 countries. A few weeks ago, we ran a similar program for ISVs on the eServer iSeries with impressive results.

You may have heard that last week we announced a new mainframe dedicated exclusively to Linux, code named Raptor. Incidentally, after Sun started running nasty dinosaur ads, we changed all the code names for our mainframe servers to be meat-eating dinosaurs.

This Raptor announcement is significant because we are continuing to adapt the platform to meet the needs of the open source opportunity, but also because this system requires virtually no mainframe skills, so an entirely new group of customers can use it. We also announced the IBM eServer iSeries exclusively for Linux, to bring the same mainframe-class capabilities to middle market customers. You can see them in the solution center if you’re interested.

Clearly, Linux products are important to us -- but we realize that their viability rests on the health of the Linux community. For that reason, we have made and continue to make contributions to the open community.

Through our Linux Technology Centers, IBM has more than 250 developers dedicated to making open source contributions to Linux as peers of the community. But these 250 people are part of a small army 20 times that size -- around 5000 people -- throughout IBM as a whole, contributing to Linux -- working in porting centers, working in research, working on our middleware and our services for Linux, working in other labs. And that doesn’t count the people tinkering with it at home on their own time.

And our contributions are being accepted. As you know, the Linux community is a technical meritocracy. To get into the community, you have to earn the right - and our people in labs and research are doing just that.

We also support the open community in other ways. For instance, in November, we launched a program called Eclipse, through which, in partnership with 150 ISVs including Red Hat, Borland and Computer Associates, we donated $40 million worth of open source tools to the new eclipse.org community. The project was designed to give developers the tools they need to create system neutral applications.

As pervasive as Linux is becoming, it is, clearly, part of a greater whole. There are other clear examples of how the open movement is affecting the future. You can see it in the evolution of Grid computing and in the powerful role the Global Grid Forum and the Globus organization play in setting the standards everyone will adopt as they roll out these grids.

Early Grids IBM is building, for example, are being implemented for the aggregation of resources among scientific and technical groups. The teragrid at the NCSA and the National Science Foundation is an example. This 13 teraflop system housed in four distant locations is made up of Linux supercomputing clusters and will be used for life sciences and climate modeling.

Another project underway is the North Carolina Genetics and Bioinformatics Consortium Grid, which will allow thousands of researchers and teachers access to number-crunching power and data for genomic analysis and drug development.

These Grids, and others like them, are designed to provide massive computational power for the advancement of science and research, but there are business Grid applications as well.

IBM sponsors a customer council made up of different kinds of businesses from around the world, called the Advanced e-business Council. Through this group, we engage with leading technical people from our customers' organizations to learn what their most pressing issues are, and work together on solving them. These are businesses focused on business issues - things like availability, reliability, scalability on demand, and end-to-end systems management.

The thing is, the open protocols you need to support scientific kinds of workloads over a grid, and those you need to deliver commercial distributed computing with the qualities of support required by businesses, such as reliability and availability, among others, are very similar.

In 1999, Ian Foster and Carl Kesselman published the original paper on Grid, called The Anatomy of the Grid. At the time, our Advanced e-business Council, or ABC, under the co-leadership of an IBM Fellow named Jeff Nick, was working on solving business issues related to commercial distributed computing and virtualization. When Jeff read the Foster and Kesselman work, he realized they were all working on the same problems. There’s commonality in the requirements to distribute work across all of these environments, whether for scientific workloads or business applications. Therefore the technology solution and protocols could be the same.

Jeff Nick contacted Foster and Kesselman to engage in collaboration. As a result, the next version of the grid paper, called Open Grid services Architecture, to be released soon, is authored by Ian Foster, Carl Kesselman, Steven Tuecke -- and Jeff Nick.

Grid computing establishes the fundamental importance of openness. There's no way you're going to deploy a 1,000-computer environment across multiple locations - universities, businesses or hosting centers - and mandate that the implementation of the infrastructure will be the same on every node. These are, and will always be, heterogeneous environments.

Together with the Global Grid Forum and with Globus, next month we expect to release a new set of specifications for common protocols that will help deliver the kinds of support businesses need in order to operate over the Grid.

Already, though, we are starting to see Grids in place designed for business. The olympics.com infrastructure from the 2000 games in Sydney is an example. IBM was the technology provider for the Olympics and for the official web site. The infrastructure we built to support the site was a Grid, distributed over the United States and Australia.

When we built it, we were committed to delivering a system that would stay up, no matter what. In spite of more than a million malicious attacks, the system never went down. Work was shifted automatically around the Grid as nodes were compromised, and we maintained 100% reliability.

There will be other ramifications for business as well. When computing resources - everything from processors to I/O to storage and applications - can be purchased as a utility service, computing will have much more flexibility, in terms of choice in technology, the way technology is deployed, and the way it is acquired.

Through our work with customers in the Advanced e-business Council, we have learned two fundamentally important things. The first is that the e-business infrastructure is a heterogeneous environment, and it always will be. People have tried and failed to build architectures around a single operating system. They failed because what customers want is to use the best of what’s new, and connect it to the stuff they have, no matter what that is.

Clearly, the infrastructure will continue to be made up of different kinds of servers - data/transaction servers, Web application servers, and appliances or edge servers, with many different kinds of architectures. As new technologies develop, we’re starting to see super dense technologies like blade servers and servers-on-a-card. In the not so distant future, universal chassis systems will allow for racks of hot-swappable devices, with different processors -- Power, Intel, firewalls, caching blades -- to be seamlessly integrated with common power and cooling in the chassis. In this kind of environment, openness isn't just an asset -- it’s critical.

The second lesson we learned from the ABC council and other research is that increasingly complex systems must be able to manage themselves. It is critically important to take the kinds of reliability technology that IBM has developed in our high-end systems -- things like Intelligent Resource Director in the mainframe, which automatically reallocates processing power as workload demands, and software rejuvenation in our Intel servers, which allows systems to run through system errors -- and deliver them not only in individual systems, but openly, over the network, to the entire infrastructure.

Through a project called eLiza that you may have read about in the press, we are working to deliver self-managing, self-optimizing, self-healing technologies in our systems and throughout the underpinnings of the global infrastructure.

Ultimately, we are working to build what we call autonomic systems -- systems which manage the basic workings of existence much as the human body handles things like heart rate and body temperature. Just as you don’t have to think about making your heart beat, systems administrators won’t have to think about rerouting work around failed components.

We see autonomic computing, which will be a requirement of the universal Grid, is one of the grand challenges of the IT industry.

I strongly believe that the Internet and technologies of the open movement, and the Grid, are the future of computing. I also believe the reason for this evolution is the free nature of these technologies. The barriers to entry are being abolished, and unfettered collaboration and innovation are the result. We have unquestionably entered the era of the open platform.

Clearly, Linux and other technologies of the open movement are changing the IT and business environments, as well as the kinds of communities people around the world are able to form. It is also no exaggeration to say that Linux has started to change peoples’ lives.

A Grid system IBM is building with the University of Pennsylvania is an example. The Grid, built of Linux clusters, will link geographically dispersed hospitals to maintain mammographic data, helping doctors treat patients over their entire lifetimes. Eventually, it may lead to new kinds of treatment.

Linux is also bridging the chasm to the Internet age for low income people. "Popular Computers" in Brazil come running Linux for $250 each. In India, $200 pocket-sized Linux "Simputers" powered by AAA batteries read information aloud so even illiterate people can tap into the Internet.

Government implementations are another example. For reasons of national security and national pride, government officials in countries including France, Mexico and Germany are increasingly adopting Linux. In China, they’ve launched something called Red Flag Linux to retain more control over their systems. The Chinese don’t want to build an economy entirely on Western technology. Eric Raymond, a recognized leader of the open movement, has been quoted a saying that a lot of governments simply don’t want to place the fate of their IT infrastructures in the hands of a single software company.

There are also examples of school programs in countries like India and Mexico where Linux is bringing education closer to adults as well as children. This kind of world altering innovation is what you and your colleagues are driving.

So, what will this new world of innovation mean to you going forward? It means, simply, that you and your colleagues, individually and as part of the greater open community, will be able to make an impact - on IT and on the world.

Back in August, IBM launched something called the Linux Scholars Program. The idea was to open a contest to university students around the world, offering an IBM ThinkPad to the students who submitted the 25 most interesting Linux projects. As part of the deal, the university with the highest average score of entries would win a Linux-enabled zSeries mainframe.

Demand was far, far greater than we had anticipated. In total, there were 1,462 students from 64 countries around the world representing 669 universities who wanted to compete. We narrowed the entries down to 25 winners, and we’re pleased to announce that Clarkson University, with 3 students in the top 25, will be awarded a mainframe Linux system.

Among the highest ranked entries was a submission from a German student named Helmut Gantzler at the University of Edinburgh, working on 3D modeling. In addition, a student named Bruno Silva and some teammates from Portugal introduced a Linux-based robot that can see and perform tasks, and named it Cyclops. And a Finnish woman, Mika Prikonen, submitted a project designed to debug applications involved in maintaining data integrity.

And this was just one little contest hosted by IBM. This doesn’t begin to illustrate the breadth and depth of innovation underway in your community.

The point here is that you and your colleagues are innovating in fundamentally new ways, forging a new direction for the industry.

Linux has proven that the open development model works. The result is an operating system that is already changing the culture of computing, and will go on to change the culture of business.

Open technologies further innovation and creativity; they foster business, they create wealth -- and they narrow the divide between developed and underdeveloped nations and people. The barrier to participation, contribution and benefit becomes one of merit rather than one of access.

The computing eras that have gone before were defined by proprietary kinds of technology: mainframes, minicomputers, UNIX, PCs. This era is defined by the technologies of the open movement.

We are confident that we made the right choice and investments in Linux and other technologies of the open movement. I believe these technologies represent a fundamental discontinuity with the past, a break with the very idea of era of proprietary platforms and control points.

You and your colleagues in the open movement are launching a new direction for the IT industry. We in IBM are proud and privileged to be part of it with you.

Open is the world’s new direction.

Copyright 2002