If one knows how to read the signs, cataclysmic events rarely happen without warning. Anyone watching the enthusiasm for Linux being displayed by IT heavyweights like IBM, HP, Sun, Oracle and SAP could not fail to feel that their efforts to give it mainstream enterprise respectability have to have an impact. Add to this the enthusiasm of the Open Source movement and a zillion university students to whom Linux is as natural as breathing and you have a bandwagon that is definitely rolling.
Some want to squash Microsoft. Others love the idea of Open Source and hate proprietary code. Some see a huge commercial opportunity looming. The motive doesn’t matter. What unites both the Linux and Open Source community and a substantial body of the top IT players, is a determination to make Linux mainstream – which means getting it into the enterprise. Not just into the edge of the enterprise, where it is today, but right into the heart of the data centre.
As the Linux flood washes in, it will, of course, also be washing out any number of proprietary Unix installations and Windows implementations, but that’s the way things go.
Will Linux “rid the world” of Microsoft, as some Linux enthusiasts hope? Probably not. Even though Linux source code is free to all, while Microsoft’s licensing strategy is viewed by some as a questionable tax on business, the investment corporates have in the Wintel (Windows/Intel) environment is too big to be killed easily.
What Linux is very likely to do in the eyes of even mainstream IBM and Hewlett Packard folk, is to unify the Unix community around itself over the course of the next few years. If that happens, the gravitational pull on independent software vendors (ISVs) is going to reach black hole proportions.
Already, as Mark Cathcart, corporate technology strategist at IBM’s New York office points out, Linux runs on everything from TV set top boxes to Intel servers and IBM mainframes. For its part, IBM now runs Linux on its entire range of systems, from its Intel servers, through the Power PC platforms using IBM’s Power PC chips (the RS6000 hardware, now renamed as the iSeries) and on up to its mainframe processors (now called the zSeries). Currently Linux runs on some 21 different chip sets and this cross platform portability is a good part of its strength, the other being the fact that it is Open Source – but more on that in a moment.
For an applications vendor, the idea of being able to write one version of an application and have it run on “everything” is hugely attractive. Per Tejs Knudsen, CEO of the enterprise applications systems house Maconomy, for example, which has just ported its entire applications range to Linux, points out that there is a huge overhead for his company in maintaining multiple versions of its source code for each target operating system. “It would be great to be able to write just one version,” he says.
Of course, Knudsen does not expect Linux to become the only enterprise operating system any time soon. The port has been done simply because the timing now seems right, he says. Clients and potential clients are not so much buying Linux versions right now as checking to see that a version will be available if they decide to go down this route. Six months ago, the question didn’t arise. He says: “We may not win any additional business right now by having a Linux offering, but at least we won’t lose any business either!”
IBM’s Cathcart, however, argues that Linux is already far more deeply seated in the enterprise than many people realise. It’s running web servers, file and print servers and databases and as it continues to prove its durability and resilience in these areas, its claims on other areas will grow. “Enterprise Linux makes sense from a number of perspectives and it will become more and more important in the enterprise space. We are seeing significant interest in it and we are making very significant investments to deliver on that interest.”
By way of an example, Cathcart points to a proposal IBM received recently from a large German financial organisation that wants to run a key Java-based Internet banking framework on Linux, using an IBM zSeries mainframe.
Questions over Linux’s ability to scale are waved aside by Cathcart and others. It is a fact that Linux can’t at present scale beyond four-way or at most eight-way SMP severs, but he points out that what one is talking about here is an eight-way server running a single image of the operating system.
There may be one or two applications somewhere in the world where one needs to run a machine with 256 processors, all running a single image of the operating system, but these are the rare exceptions, he notes. The problem with gigantic, multi-processor single image machines is that it becomes very tricky to carry out any kind of upgrade without taking the entire system off line. Most people who want vast computing power tend to run a large number of two-way processors, with each pair running a separate image of the operating system. Used like this, Linux can and does scale to supercomputer levels very well, and it is also very much easier to manage, since it can be taken down in chunks, while the machine as a whole stays up and running.
Mark Yates, strategic alliance manager at the German-based Linux distributor, SuSE, reckons that the consultancy side of Linux is growing in leaps and bounds. SuSE has its own consultancy service he says, and works closely with smaller consultancies. “Germany is probably the biggest enterprise Linux market in the world at the moment. Our consultants are in there advising on architectures and solutions. Deutsche Bank, Daimler Chrysler and many others are all there running SuSE Linux,” he says.
Colin Tenwick, vice president of international operations at the US and European Linux distributor, Red Hat, is adamant that Linux represents a huge opportunity for the consultancy community. “Demands on IT are increasing, but budgets are shrinking. Linux has an unbeatable story to tell when it comes to price and performance, so the global slowdown is giving real traction to Linux in the enterprise,” he says.
Red Hat recently bought PTI, a 220-person consultancy to build up its professional services capabilities. “We are now getting substantial demand to become involved in the architectural build out and project management of large scale enterprise Linux implementations. Moreover, since this is too big for one company to manage, we are now actively building up our partnering relationship with a network of consultancies, providing them with Red Hat certification programmes for their people,” he says.
An old argument against Linux in the enterprise, that one still hears trotted out from time to time, is the objection that an Open Source product like Linux, which does not have a single, proprietary owner with deep pockets for R&D, is not going to be able to grow as well as a proprietary operating system such as Windows 2000. This argument flies in the face of so many facts that it is now a real mark of either its proponent’s ignorance or malice.
As Linux enthusiasts point out, the operating system has matured far faster than any other OS on the planet. The not-so-small army of un-paid enthusiastic Linux developers have kept polishing and adding to the code. New and improved versions of the kernel keep being released at respectably rapid intervals and Linux has also benefitted substantially from mainstream IT players like Sun, IBM and HP, all of which have added important elements to the Linux code base.
One of the more dramatic signs of the rapprochement of mainstream IT players and the Open Source community was HP’s decision, in December 2000, to recruit Bruce Perens, one of the key founders of the Open Source movement, as one of its top five executives, and to appoint him to be its Linux strategist and liaison specialist with the Open Source movement. In fact he has just finished writing HP’s Open Source corporate policy.
Says Perens: “If anyone had told me back in 1996 when we’d just announced the Open Source movement, that I’d be setting HP’s corporate policy on Open Source within five years, I’d have dismissed it as a flight of fancy.” For Perens, the fact that there are more than 20 Linux distributors around the world (“including a number I’ve scarcely heard of”) should not be viewed by corporates as a weakness.
“I read an article the other day in an IT journal saying Linux needed to consolidate to be taken seriously – it’s nice that they’re writing about Linux now in mainstream IT journals but that kind of comment shows that they still don’t really get it,” he says. Multiplicity is Linux’s life blood, he argues, particularly since it is now fully complemented by a formalised standards body (to be found at www.freestandards.org). “Enterprises did not take Linux seriously until standards emerged. They were afraid it would break into a million pieces, like Unix. Now you can take a standard Linux application and run it across all kinds of platforms. Anything that helps Linux retain its diversity while allowing enterprises to address a single standardised operating platform is important,” he says.
According to Perens, although the big consultancies are not yet visibly engaged in major Linux projects, enterprise Linux is off and running. Moreover, when the big consultancies do move into the Linux space, he says, they’ll find they won’t have to pause while they recruit Linux specialists – “what they’ll find is that a good proportion of their software engineers are already running Linux at home,” he says. In fact, Perens – and many others in the Open Source movement – point out that there is already a huge “underground” penetration of Linux into the enterprise space in companies where there is no management sanction for it.
“Technical folk are putting up Linux web servers, file and print servers all over the place. They love them because they just run forever. When management stumble across these, if they have any sense, they either walk away or else they do a little digging, find out how well the thing’s been running, and then start integrating it into mainstream corporate IT policy,” he notes.