Linux in the Enterprise Closer Than You Think
By: Brian Proffitt
Thursday, December 12, 2002 01:50:18 PM EST
URL: http://www.linuxplanet.com/linuxplanet/opinions/4581/1/
Meeting the Enterprise
There has been a marked increase in the word "enterprise" within the Linux community of late, and it has very little to do with the latest iteration of Star Trek. It's a catchy little phrase that encompasses all that is corporate in Western society. If something is big, it's the enterprise. If it has a lot of money, it's the enterprise.
Try to pin it down much farther than that and you can get hung up. If big corporations are enterprise, then how would something like the United States Postal Service fit into that definition? It's certainly big, but it's also a government-run organization. If they started using Linux, could the USPS be embraced within the "enterprise Linux" framework?
Probably.
I have my own definition for what the enterprise is, which I use in making judgments about where certain stories on Linux Today and LinuxPlanet will be placed. The enterprise is anything that either uses a lot of resources to get something done, or needs a lot of resources to get something done.
When I was in Boston last week at the Enterprise Linux Forum (ELF), I found a lot of people from organizations that would fit into this definition. And all of them were looking for ways to start getting Linux into their IT schemas. Now, to be fair, many of them were using Linux already in one way or another. But now the time had come to start using Linux in ways that would not only allow them to do their work but would also let them do their work better.
What the Enterprise Needs
There have been quite a few articles on the topic of what the enterprise needs to properly implement Linux of late. Quite a bit of it is fall-out from that aforementioned forum. The press, I will freely admit, love such affairs because they are usually great places to get a lot of stories in a short amount of time. That, and the free food.
I have watched with some bemusement the overall trend of the articles that are coming out on Linux in the enterprise, and not just the ones from ELF. What Linux needs, these articles intone, is more applications to be successful. If only there were more applications for Linux, pundits and corporate IT people have said, then it could not help but be more successful.
Let's be clear here: they're not talking about just any old application. Given the sheer number of apps and utilities that have been built for the UNIX/Linux space, such requests seem rather absurd. No, what these people want are neatly wrapped pre-packaged applications that will install easily and run without a hitch on their systems with little to no configuration time involved.
The range of these applications is great--from huge databases to little utilities that will tell you what the temperature is outside. Office suites are certainly in there, though there is more concern for what's going on with the servers than with the client. Grid computing is in there as well, ramping up to true supercomputer-speed computing efforts that have already caught the eye of the financial services industry.
Linux can handle all of these things now, of course, but for IT managers that have put all their eggs into the Microsoft basket, the transition to these applications is perceived to be too hard.
This could be argued against. But this is another argument for another time. If there is indeed a real need for applications, however, then there is evidence to suggest that independent software vendors are responding to the call.
Already we have seen companies like Oracle, IBM, Veritas, Steeleye, and SAP jumping into developing applications for the Linux space. According to a study from IDC, the biggest growth in Linux right now is in the area of custom application development. Linux is still big in Web services, according to Jean Bozman, VP of server research at IDC, as well as in security and collaborative workloads. But custom applications development is growing fast.
Are applications going to be the panacea for Linux in the enterprise? Perhaps. When I spoke to Bozman earlier this week, I asked her to highlight the differences between Linux server adoption now and Windows server adoption in the mid- to late-1990s. She related that even though Windows NT was out in 1993, it wasnot until NT 4.0 came out in Augsust of 1996 that things really took off for NT in the server space. The difference, Bozman explained was that NT passed its main competitor Novell not in terms of a network operating system but as an application server.
In other words, she said, it surpassed Novell's application base and therefore became more attractive to IT shops.
There is no denying that applications are a key to success to Linux in the enterprise. But it is not the only key. There are other things the Linux community could do to accelerate the adoption rate.
What Else the Enterprise Needs
Besides applications, there is a clear need for other sucesses to get Linux adopted in the enterprise. After talking to some of these people at ELF last week, here are three of the items that came up in my many conversations.
Standards
This one was a big one, and it was heavily mentioned at the event by attendees. One of the biggest problems in migrating from one platform to another was the jump in file standards one had to make. For many at the conference, the open-source adherence to standards was a big selling point for Linux already.
What made it hard, the people I interviewed said, was the fact that outside of Linux, these standards were little used. No one was naive enough to think that this was anything but delibrate on the part of Microsoft and other proprietary software companies.
But even though Linux has open standards already, there is something it could improve in this area.
"If the Linux community could do one thing better," one gentleman told me, "it would be to advertise the hell out of the fact that open source also means open standards. That message has got to get out more."
This sentiment was confirmed by others. Open source was easy enough to grasp: it means open access to the source code. Standards, though, are only implied to be open. IT managers, sick of being locked into forced product upgrades when proprietary file formats are upgraded, would love nothing more than dedicate their files to open standards.
So, for those of you investigating Linux for the first time, let me be first to let you know: many of the file format standards used by Linux applications are open.
Drivers
Linux needs more drivers, of that there is no question. All of the varied and sundry pieces of hardware out there in the corporate world are screaming for driver creation.
In one conversation I had, a senior research computer scientist from the United States Courts summed up the problem with driver creation quite neatly.
"Drivers," Robert Borochoff said, "are boring."
While writing a driver that worked well was a relatively simple thing, the bulk of the work in drivers is the tedious coding of error handlers within driver, Borochoff explained.
But it is exactly this kind of tedious work that is needed right now, he said.
He has a pretty good point, too. Driver work is not fun, and there's very little glory in it. Not to mention the incessant patent and copyright problems that are always cropping up.
This latter may soon be less of an issue--independent hardware vendors are standing on the brink of Linux support, waiting for just a little bit more adoption before jumping in. Once their decision to support Linux is made, then there will be a clear market for dedicated, meticulous developers to start integrating drivers with Linux.
Education
There are two kinds of education surrounding Linux. People not involved with Linux clearly need to be educated about just what Linux is all about. I can think of several DC-based pundits that could use a real strong education right off the top of my head. Not to mention some analysts.
No one knows everything about Linux, but there is a clear need for Linux users to deliver a single, unified message about what Linux is, what is the difference between free software and open-source software, and why the GPL license is not viral.
It's not that these are hard questions to answer--it's just that there is no central organization that is handling this kind of knowledge. That has been Microsoft's one true strength: the ability to deliver its message from a single, unified source. Many may not like the message that they are sending, but you have to at least give them that point.
Without such a central organization, then it's important for people knowledgable about Linux to get the message out on their own. Deeds, such as setting up a Linux-based network for a local charity or small business, are always better than words. But if words are to be used, it's no longer acceptable in any way to toss out "RTFM." If you want to be cool and aloof, that's your problem--don't make it Linux's.
The second kind of education is the more formal kind. People need to get trained up in Linux particulars as soon as possible. I spoke with Evan Leibovitch in Boston and this is something that the Linux Professional Institute is working hard to achieve with their top-of-the-line Linux certification programs.
LPI has recently overtaken the Red Hat certfication program in terms of most desired certification and after meeting Leibovitch, I can see why. This is an organization that wants to get as many people formally trained in Linux as possible--but not at the expense of stuffing the numbers.
For instance, Leibovitch is very proud of the fact that the LPI certifications have a high fail rate. When I first heard that, I was glad my kids were 800 miles away. But as he continued, I realized that this was no sadistic ogre before me. The high fail rate is good because everyone in the industry will know that the people who do pass LPI certification exams are clearly going to know their stuff.
LPI is going to be expanding its efforts around the world in the months to come--a story for another day. But their presence today is a strong plus to solving the problem of formal Linux education.
What the Enterprise Already Has
You may have noticed a theme in all of the needs that were raised by the attendees of the conference--they are needs that already have solutions of a sort.
Applications and drivers are being built, standards are already here, and education (at least on the formal side) is coming along rather nicely.
So what is the problem?
It is, as usual, perception.
Linux is perceived as being too hard. Too decentralized. Too (fill in the FUD here). Both Russell Pavlicek and John "maddog" Hall both gave good talks on this at the ELF. The message about Linux has too long been hidden by misinformation or faulty data.
The problems that Linux used to have are still being perpetuated as gospel by the uninformed. And, without curbs from those who do know the right answers, these messages are heard by a willing public.
On the way home from the conference, I arrived early at Logan Airport and settled into my gate for a long wait, which was fine by me, since I had an article to write. I opened up my iBook and booted up SuSE to get some writing done. There is a problem with iBooks--they have that silly glowing apple that attracts too much attention.
Sure enough, a businessman sitting across from me noticed it and asked me how I liked the new OS X.
"It's okay," I replied, "but I'm running Linux on this now."
Whereupon I had a (decidedly one-way) discussion with a guy on his way to Allentown, Pennsylvania about everything that was wrong with Linux. It was all stuff that he'd heard around the office from his IT crew or from watching CNN.
He was a fast-talker, so it was difficult to get a word in early in the discussion. Eventually, I was able to show him the desktop and explain to him all the things I could do with Linux. He came away more than a little surprised. After all, here I was, actually using the thing without any concessions to quality of work.
Linux users see this all of the time. Someone does not know about Linux and assumes the worst is true about it. Until they actually see it in action.
And that's what Linux needs to get in the enterprise: it needs to be seen. The message needs to be heard.
Now that that process has begun, it's only a matter of time.