It has become a tool like any other.
I have seen a lot of different network setups in my travels and doings. And what I’ve seen has made me think about what I read today in an article speculating about the future of Linux. The article considers the idea that Linux is no longer considered a “winner”. It talks about what the word “winner” can be taken to mean, but the idea I get from the article is that people don’t think of Linux as they did in the past.
When it was first getting popular in the beginning of the middle of the 1990s, many people saw Linux as a savior. Until then, we were pretty limited when it came to what systems we could use on our commodity x86 servers. We could either go with a MS-DOS solution, or a SCO OpenServer or UnixWare solution. There were other systems available, of course, but those were the main ones we were dealing with. Many people would avoid BSD-based systems due to the uncertainty caused by the legal troubles, for instance. So when Linux became available, and it proved to be usable, I know many of us were quite happy. Here we had a powerful UNIX-like system at our disposal, with full source code available to us. Combined with the GNU utilities, we were basically set.
I have to agree, that is not the mindset we have today. Linux has become entrenched in many organizations. It’s not really seen as the savior it once was. This is especially true when we consider all of the other developments that have happened since the early and mid 1990s. The major two have been FreeBSD and OpenSolaris.
FreeBSD, and to a lesser extent NetBSD and OpenBSD, offered us all a really great alternative to Linux. BSD/OS was another option, but most people I know just ended up choosing FreeBSD or NetBSD. I know that it can become a heated dispute between the proponents of the *BSDs and the proponents of Linux. But what I found from working in the trenches is that when it comes to servers, I’d rather be running FreeBSD than Linux. This is mainly due to, in my experience, what was higher reliability and performance from FreeBSD. While that has changed somewhat since then, I know several system administrators who much prefer the coherency of FreeBSD to that of the major Linux distributions.
So in my opinion, FreeBSD reduced what was special and unique about Linux. It provided a free, open source x86 operating system suitable for server use. The main reason I think Linux may have remained ahead is media hype and corporate support from industry giants like IBM, as well as better hardware support.
More recently, the creation of the OpenSolaris project has likely had somewhat of an impact as well. Many of us who have worked in the field for years have become very familiar with Sun’s Solaris. We know that it is often an excellent platform to use on servers, and it also makes a damn fine workstation OS. With it now being available as open source software, the unique aspects of Linux have further been diminished. I know several administrators who had years of Solaris experience on Sun hardware, and who are now moving their infrastructure from Linux to Solaris on x86 hardware.
At this point, Linux may have lost a lot of what made it unique. It now has to deal with some serious competitors in the form of the BSDs and OpenSolaris. For many situations today, it doesn’t matter much which one you choose. In many cases an infrastructure can be built upon any of those systems, and it will likely perform well enough. It becomes more a matter of the experience of the system architects and administrators, and what platform they would prefer to work with. Even then, many are flexible. They know Linux as well as they know FreeBSD or Solaris.
These days, it doesn’t surprise me if many people don’t see Linux in some glorious light, as many did a decade ago. That is because it has become such an integral part of our toolbox. We know where it can and should be used, and so we use it. Just as we don’t think twice about hammering in a nail with a hammer, we don’t think twice about using Linux to solve a difficult server task.
January 9th, 2007 at 7:28 pm
We started out with Solaris in this lab. The guys wanted to move to linux (which you’ll notice I didn’t capitalize). They thought you could just bring in another OS like linux, plug it in, and replace the pricey Sun hardware.
“Because everyone else is doing it and the people I talked to said it worked fine”
I just bought a used 440R for code that they can’t port or applications not available on linux. And I warned them about 32 bit vice 64 bit / little endian vice big endian — in one ear and out the other until it was too late.
We “migrated” to linux 2 1/2 years ago but everyone still requires an account on the Solaris box to get their work done. Go figure.