Linux Comes of Age

In the last year, the Linux story has changed from the adventures of an interesting band of eccentric nerds to an operating system that draws the increasing interest of corporate IT. Linux is actually becoming respectable!

According to IDC numbers recently quoted in the Wall Street Journal (April 9, 2001), Linux has moved up from 16% of the server market in 1998 and 24% in 1999 to 27% in 2000. (For comparison purposes, Microsoft W2000 has a 41% market share, up from 38% in 1999.)

Such respectability necessarily brings change, some of it long desired by Linux aficionados (and probably needed), some of it feared by Linux purists.

All of this has taken place in a declining stock market, where newly created Linux companies have been treated unkindly, loosing significant market valuation and, with that, their ability to use their market capitalization to easily expand through acquisition. Instead, they must take the harder and longer route of finding partners and building customer relationships and revenues, all in a limited resource environment.

While individual technical users – and webmasters buying cheap, reliable servers – continue to be strong buyers of Linux, the up tick in corporate activity is clearly visible. On the Buyer side, a variety of businesses are noticing the cost savings, stability, and reliability of Linux, as well as its enhanced features, the interesting aspects of the Open Source development model, and the availability of increasing amounts of business software.

Smaller businesses typically buy Linux through systems vendors such as VA Linux or VAR’s who specialize in particular vertical markets, with expertise in the Linux platform. We’d expect to see more of that now that the Caldera/SCO deal has closed and SCO is getting out of its traditional Unix/Small Business market, leaving this playing field to Caldera. Also, many Intel server vendors, including IBM, Hewlett-Packard, and Compaq, but also including Dell and Gateway, now offer Linux support on their hardware, extending the Linux market to both their direct customers and through their VAR partner channels.

Fortunately, the entrance of many large, stable companies, especially traditional systems and software vendors including IBM and HP have permitted corporate confidence in the future of Linux to continue its rapid growth.

It does mean that much of this growth has benefited large, well-known companies rather than the smaller, newer, more Linux specific companies.

Corporations are now routinely considering Linux for all sorts of server applications. We find this story can best be told by looking at what appears to be the broadest current strategy, that of IBM.

In fact, IBM has become a major factor in the Linux market. For IBM, Linux solves (or moves towards solutions) in several areas:

· It supports cross-platform application development. IBM customers are very likely to have a heterogeneous environment, including departmental servers (NT and UNIX), lingering OS/2 servers, AS/400’s, and mainframes. Linux operating systems can run across all of these environments, IBM offers Linux-supported middleware for all of them, and applications that run on Linux, run on Linux (with the usual caveats for systems eccentricities and limitations).

· It’s a competitive solution that it’s unlikely Microsoft or Sun will try to meet. That gives IBM competitive advantage on both its hardware platforms (versus Sun in the UNIX market) and in the middleware market (versus both Microsoft and Sun’s iPlanet group).

IBM believes that Linux crossed the threshold and moved into corporate applications in 2000 for a variety of good reasons.

· Saving money is now a more important issue. Linux is cheaper to acquire, essentially free, and generally viewed as cheaper to administer and support than other (read Windows) operating systems, based on its greater robustness.

· While Linux isn’t ready yet for every Enterprise application (SAP, for example), it is ready for a number of mission critical tasks. IBM delineates four areas where it sees corporate customers buying into Linux:

Many corporations go to Linux as a way of providing a low cost, very reliable, highly scalable environment. IBM says 50% of their Linux sales are in this environment. Here you will see such traditional Linux server applications as web servers, firewalls, caching servers, print and file servers, and messaging servers. This is often the Linux entry and proof point for many corporate customers.

An important area where Linux’s very low price is clearly a consideration is this one, especially in retail distribution (so far stores have been prime customers) and financial services. Insurance agencies are also an important part of this market segment.

Nearly everyone has read about the splashy Linux win in Japan, where Lawson Stores will be installing 15,000 Linux point of sale servers with remote management and support. IBM’s Global Services will supply the implementation for this large project.

Banks are also looking at Linux for branch office systems, often replacing older OS/2-based systems. IBM reports that there are pilots underway in both the U.S. and Europe. This might be a good time to mention that Linux is an international, not a U.S., phenomenon. Germany is one of the most aggressive countries in embracing Linux, helped by the SuSE Linux distribution, based in Germany and the fact that IBM’s work to bring Linux to its mainframes was done at its Boeblingen laboratories.

A surprising turn of events for some analysts has been the strong interest of IBM customers (and some new customers) in using IBM’s z-series (IBM’s newest mainframes) and soon their i-series (AS/400) servers as Linux servers for the purpose of consolidating what would otherwise be running on large numbers of smaller servers.

IBM permits a z-series customer to run Linux in a partition while continuing to run the normal mainframe operating system (and its software and data sets). Special pricing for Linux middleware is available not just from IBM, but also from Computer Associates, Compuware, and others, which can make this a very appealing idea, saving considerable money.

Currently, there are over 100 pilots running Linux in this workload consolidation mode, including application in financial services, service providers, and telcos.

Linux started on Intel and when hundreds of these systems are clustered together (IBM can currently cluster up to 1,024), it is possible to reach super computer power at relatively low cost. Universities and scientific and technological applications such as life sciences were initial customers. Financial services for risk management and portfolio optimization appears to be another good target market. Clusters targeted to specific vertical industries will be arriving later this year.

IBM has a Linux strategy for each of its hardware platforms. For the P-series (its AIX-based Unix servers), it’s a multi-part strategy. Today, IBM offers software that permit Linux applications to be recompiled to run in the AIX environment. Later this year, IBM will support Linux running in partitions on the P-series platform.

Linux itself changes on a regular basis. Linux developers can, of course, make any modifications they choose to the Open Source code. Changes are regularly offered, refined, and considered for the next update of the Linux kernel, still nurtured by its inventor, Linus Torvalds.

In January, 2001 Linux Release 2.4 was published; work on that release began in May, 1998, and this release was considered to be too long in coming.. In March, Linux developers gathered at the Kernel Developers’ Summit to discuss possible changes for the next version. A beta version (Release 2.5) is scheduled to emerge later this year and the release itself (Release 2.6), hoped for by year’s end, but is likely to emerge sometime in 2002.

The goal is to do much better, from a scheduling point of view, this time. It’s hard to meet schedules, however, when the programmers writing the code are volunteers, located anywhere, doing the work partly because they enjoy it and partly for the glory. Scheduling of new releases will clearer be an important pressure point with larger numbers of corporate clients in the picture.

In line with the corporate interest in Linux, the emphasis in the 2.6 release is on scalability, larger memory, and support for enterprise class databases as well as enhanced Internet protocols, security (a big Linux/corporate issue), and laptop support. A Webcast of presentations given at the event is available at http://www.osdn.com/conferences/kernel/.

One of the reasons it’s hard to track the Linux market is that market research firms count servers and servers are not necessarily a good measure of either revenue or processing power. For example (numbers are from IDC for the 4^th quarter of 1999, similar figures for 2000 should be published momentarily),

But this fails to take into account the fact that IBM may include mainframe servers in its numbers (and surely will in the 2000 numbers). Such servers may take the place of hundreds of individual Intel or UNIX servers and, of course, sell for much higher prices. Counting up server noses is one way of looking at a complex market, but it may not tell you very much about what’s really going on if many of the large Buyers are moving to larger scale Linux solutions.

IBM is not, of course, alone in this market. Many vendors of Intel servers to the Linux market offer high-end cluster configurations. For example, Linux specialty house VA Linux has recently announced new servers for its high-end cluster configurations, optimized for dense implementations in applications like web server farms and at telcos.

To further muddy the waters, most IBM Linux servers, regardless of size, are sold with substantial quantities of IBM middleware – web servers, systems management, messaging and collaboration, data bases, etc., etc. Most other server vendor’s partners with software vendors. They take a share in the profits of middleware offerings shipped with their servers, but only a share.

Linux has been mainly a server operating system. Of course, there are Linux aficionados who have the enthusiasm and skills to use Linux on the desktop, but so far they have been only a small part of the overall Linux market. On the desktop, where the user is directly in contact with the operating system and its user interface, Linux was perceived as too demanding for most mere mortals. In addition, very few popular desktop applications have been ported to the Linux environment, so as long as the focus of users is running software on their own desktop computers, this has been a significant barrier.

Desktop application development focuses on several lines of attack: existing vendors port applications; new vendors create applications for the Linux market particularly or add Linux to their list of supported or to be supported platforms as they plan their marketing strategy. Internal IT programmers and Systems Integrators and VAR’s are also adding to the body of desktop Linux applications which will eventually find their way to market.

Key to this thinking is a distribution strategy. While Linux itself is an Open Source play and is distributed for free as a download (or for a small fee for providing an integrated distribution and its documentation), applications are generally not free. They may, however, in the spirit of the Linux market, be priced differently than those for other platforms. Sun, for example, distributes the StarOffice Suite for free.

Much of the Linux desktop software market operates on the web as downloads, efficient and economical. But corporate buyers may want something more controlled and better supported, so as the corporate market grows, on both desktops and servers, we’d expect to see higher priced and better (perhaps more formally is a more politically correct word here) supported options to appear.

There is also a growing interest in a Linux component market (just as there is already an interest in a web-based Java and XML component market). Some of this occurs via Linux sites, but also through intellectual property marketplaces and programmer marketplaces like HotDispatch. This is still in its infancy, but we’d expect corporate users to find it especially useful for outsourcing small jobs for which they lack specific skills.

Linux is all about Standards and the political correctness of the Open Source lifestyle. Corporations like the upside here – the promise of things working together on an ongoing basis and the existence of a vast community of smart people who will help you out – but may find the downside (from their point of view) hard to take. This is a world in which corporate users must settle for less control. For example:

· It’s fairly typical in the usual vendor/customer relationship for big customers to have lots of clout. That means that big customers can get features added to operating systems, middleware, and applications, sometimes to the detriment of the product design or the overall cost of future support. This can be especially true early in a market or when several of your big customers come from a single market segment and have similar special needs.

In the Linux world, unpaid developers and their spiritual leader, Linus Torvalds, make decisions, based on technical issues and their sense of priority (which often has more to do with what important members of the group wants to work on next than with the demands of a commercial market.

We expect that as large companies enter the Linux market, as both customers and vendors, this will inevitably change. Customers understand the uses of influence and large vendors expect to need to keep their largest customers happy.

· Small vendors lack both resources and cash flow. Therefore, it’s easy to change the course of history by coming in and offering them business (revenue and cash flow) in return for spinning the course of product development in a favored direction. Do that a few times and a company may end up with a completely different focus – or none at all, but it might end up with customers and profits. It’s very seductive.

Open source developers don’t belong to anyone (in a general way) other than themselves. They can choose to continue working on interfaces or performance or whatever they find interesting and important – even if that’s not what customers are interested in. In that case, customers will wonder off, but they weren’t looking for customers anyway, but rather converts to their religion.

Large vendors have plenty of resources (although not infinite ones) and can choose to use some to satisfy early customers (they often seek out particular kinds of customers to enter into partnership agreements) and still maintain a reasonable pace toward their intended goal. Their problem is more often that they aren’t as flexible as they should be in responding to changes in the market so they fail to apply their resources in the best possible way – making room for at least a few small, new companies to succeed in each round.

A real issue for Linux, as big vendors and big customers enter the market is whether the graceful ways of Open Source development can survive the deep pockets of pressured development schedules vying for a scarce people resource. If they manage to pull enough skilled Linux programmers out of the Open Source pool, we may need a new development model. Big companies might not mind the return of control, but they certainly wouldn’t like the change in the pricing model.

So who, then, is buying Linux in the big company market? No one is surprised to hear that IBM mainframe customers have discovered that running Linux on a mainframe can be a very convenient solution, providing a familiar and robust environment. But IBM reports that there are customers putting Linux on mainframes that never had an IBM mainframe at all or never used it in this part of their company. Such customers are likely to be using these big Linux servers for virtualization, to be able to run thousands of simultaneous Linux session on a single big server. The ability to run both Linux (new) applications and native OS (old) applications at the same time is also viewed as a big win.

· Canadian ASP coreFusion uses Netfinity servers and Linux to provide a robust environment for providing Domino web hosting. The ability to offer multiple virtual servers on a single servers allows coreFusion to also offer its services to smaller customers who would not otherwise be able to take advantage of the software. coreFusion expects to offer selected applications from the Lotus business partner community.

· ERP Central is a portal for ERP consultants. They offer ERP news, job postings, and other information, but their big “traffic builder” is a free time and expense tracking program which users can access to maintain their schedule information and submit it back to their offices from the site. Linux hosted and built on top of Websphere and dB2, the application can scale to handle over 100,000 users and organizations whose consultants use the software estimate that it saves them 75% in time savings, an average value of $500,000 per organization per year.

· At the Wimbledon Tennis Championships, Linux, dB2, and Netfinity servers make it possible to offer real-time information on scores to fans around the globe. Last year, over 914 million web hits occurred during the games, requesting scores and statistics. IBM has been working on real-time score presentation systems for a dozen years, for both Wimbledon and the Grand Slam matches, incorporating new technological advances over time.

To continue to attract customers to the Linux platform, however, IBM and other vendors will have to attract application vendors (as compared to middleware or software infrastructure vendors) to the platform. IBM plans to do this with a new test site, which allows a software developer or an internal IT department to test an application for the Linux mainframe environment. Tests for appropriateness are offered and those who pass these tests can apply to IBM for assistance in porting their applications.

If Linux succeeds in the corporate environment, it will do so by pushing to the side opportunities for Microsoft’s Windows 2000 and XP environment and Sun’s Solaris. Both of them maintain that Linux will never have the proper qualities or support for a corporate class- or service provider- class operating system, but IBM’s support for Linux and its success with corporate customers in the Linux market may make this harder to maintain.

Linux is undergoing a period of expansion. Both its traditional Open Source supporters and a variety of new partners in both new and traditional companies are now extending its features and building new Linux applications. This activity has greatly enhanced Linux’ already considerable market potential and it is possible that in Linux we are looking at the operating system that Unix might have been if only we had been able to select and support a single version of Unix.

But vendors and customers are both fickle, always looking for the Next Big Thing. It will take another year or two to tell whether Linux has staying power and continues to build significant vendor and user commitments, or whether it was simply one more chapter in the convoluted history of UNIX itself. It looks as though Linux has already gone past that point, but we await customer and application activity to tell the final story.

	*High Volume Infrastructure* · Web servers Caching servers · Print/File Messaging · Firewalls
	*Distributed Enterprise Applications, focusing on* · Retail Distribution · Financial Services
	Workload Consolidation on z- or i-series Servers
	Cluster Solutions (Multiple Intel Servers) Super Computer Power for · Universities/Scientific/Technical/Telcos · Financial Services