Blog This!   Lee Geistlinger's Web Log

Blog Home
Blog Archives
LittleGhost Home

E-mail: geistlinger AT gmail.com

Everyone loves lists Books, Movies and so on

My Online Aggregator

Today is the 10th birthday of Linux (the release of the kernel by an obscure Finnish guy named Linus T-something).

Wow. Just had the 20th birthday of the IBM PC -- the "desktop breakthrough" -- and now this.

And yesterday -- Friday, Aug. 24 -- marked the release of Microsoft's next generation OS -- Windows XP -- to OEMs (release to public still scheduled for Oct. 25, I believe).

OK, where does that leave us? I guess with some OS issues to discuss.

What follows are ruminations mainly on Linux, but the release of WinXP and the anniversary of the IBM PC will also figure as nice counter points.

• The story of Linux, of course, is really the story of open-source software at its best (collaboration, distribution, high utility with no cost). While I have slammed the open-source movement in the past -- and will continue to do so in the future -- for being a little too starry-eyed (have to make money somewhere dude....), Linux is the poster child for what open-source software (OSS) can do. Kudos to all.
  
• Let's get one thing straight: Using Linux is not a slap at MS. Using Windows is not a slap at OSS. Many people -- especially in OSS -- appear to feel that way, but this is inaccurate. Both are tools, and they may or may not have overlapping uses. MS is the "tyrant" right now because they are so big. But -- Red Hat has caused some grumblings in the OSS community over the past year. Why? Bascially, because they are the overly dominant Linux vendor. If SuSE or Mandrake had made the same moves Red Hat did, they would be congratulated, not slammed. Let's keep this in perspective, folks.
  
• Linux does compete with Windows on both the server and desktop.
  
  
  
  • Servers: Linux is a very real threat to MS on the server side, for the following reasons: Both run on relatively low-cost Intel boxes; Linux is more stable (it is; get over it...), cheaper (OK, spend $39.95 on a Red Hat disk) and has much better security (in this case, it's not so much that Linux is good as it is that Windows/IIS is really weak). Linux is also faster for most applications. Downsides to Linux on the server side are many, also, however: Linux is -- I don't care what geeks say -- very difficult (relatively) to deploy and tweak. NT/IIS is much easier for novices -- and there are a lot of novices out there. And there is still the lingering problem -- although it is fast disappearing -- of drivers and such for Linux. This is not as much an issue on the server -- just need Ethernet drivers etc -- as it is on the desktop (printers, video cards, scanners....) -- but still an issue. The one major drawback to Linux on the server, and I see this mainly at mid-sized companies, is that this creates a split environment: Need sys admins with both skills or two sets of sys admins. People costs are huge. Pay the MS license and not have to hire another worker. Much better for business. For larger companies, this is less an issue, as they are used to having a desktop help/Web admin staff split (Windows/some Unix flavor). Small companies rely on geeks to do both, who do it willingly. All things said and done, however, Linux will continue to expand on the server. This will hurt MS and UNIX companies, such as Sun. NOTE: One wildcard that I have trouble reconciling: Database server platform. Right now, MS SQL is the best product that MS has; runs only on Windows. As I've mentioned, OSS databases are doomed to live on only at small companies that run the database and Web server on the same box (stoopid!). Only Linux options left are DB2 (only the IBM-aligned will do this) or Oracle (many large companies on Oracle, but it costs a fortune). My choice is still Linux Web server (Apache or iPlanet) and MS SQL database. Split platform, yes. But nice option. If I had the money, an all *nix deployement with Oracle database.
    
  • Desktop: Again, get over it. MS will rule the desktop from at least a few years to come. As mentioned in the Server section, Linux is still too hard to configure (people have trouble with Windows; want to mess with the mess that is a Linux desktop?). And the tools just are not there, which is key. As I have mentioned before -- probably several times -- one has to use Word and Excel. These a business basics. Until Linux desktops have apps that can read and write these -- and work in the same way -- forget the desktop for business users, which is the overwhelming use of desktops. Home users? Remember, a lot of homeowners still buy Macs..... The driver problem is still here and more pervasive on the desktop for Linux, but it is getting much better quickly. Except for geeks, I cannot see any individual selecting a Linux desktop for home use unless he has the same at work (and chances of that are almost nill). Just makes no sense. Little software (is there a Linux version of AOL?), different look and feel....no, not going to happen.
    
  
  
• Let's hope Linux does not fragment like UNIX did (Solaris, AIX, FreeBSD etc.). I don't think it will, as there is now a cohesive presence of OSS leaders who will pressure to NOT have this happen, but...who knows?
  
• Linux has changed considerably in focus over the last decade. This is normal, but people don't seem to want to acknowledge this. Yes, they trumpet the inroads they have made -- clustering, multiple processors, blah blah -- but don't seem to mention what has been left behind. And it's always a trade off. When Linux came out, one of the cool things about it was you could put it on a 386. That would be a trick today. First of all, 386s came with what, a 20-80MEG hard drive. Red Hat v7.1 minimum install is much larger than that even for the server (no GUI) version. And the 2.4 kernal (RH v7.1) takes way more memory, which Red Hat at least acknowledges such in its installer program. Most 386s had -- tops -- 64M RAM (I think mine had FOUR). RH v7.1 wants about 256. Ouch. But as I said, tradeoffs. Much better, more flexible, easier-to-install systems. And memory -- HD and RAM -- are way cheap today. But realistically, you'll need a fairly recent Pentium to run Linux at home today. My Pentium Pro box is an old one; to run v7.1, I'd have to add another hard drive and updated BIOS (again, too fucking complicated...). And it's a Digital box; my guess is that I can't even get an BIOS update for this fossil. RH v6.2 might be the last installation that takes advantage of old equipment.
  
• Linux is a good, solid OS, and that has helped it survive. However, a good product does not guarantee its survival: BE OS, NextStep, Mac OS (pre-10) and so on. Why did Linux survive where others didn't/are marginalized? I think the masterstroke was the two-fold: 1) The early Linux distros could actually run on relatively cheap and abundant Intel boxes. No, didn't need a $250,000 Sun box. Just an old 386. 2) This platform was also the platform for The Evil Empire -- Microsoft. Made running Linux that much sweeter for geeks. Wipe Windows OS off the box, install Linux and you have a fast, stable server or whatever. How cool is that to a geek? Very.
  
• Linux also survived due to the gentle yet firm guidance of Linus and Alan Cox (esp. the former). It keeps the issues low-key and almost folksy, even as Linux becomes a household word. There really has been no deviation from the original intent of Linux: A Unix for the masses, made by the masses, for free and kept open. This has, if anything, gotten better as the years have gone by. Yes, companies charge for Linux now, but that is for service/ease of installation/additional tools. Anyone can still download the Linux tarball and install for free.
  

That's a pretty impressive first decade: From a "hobby" toy to an OS that is the flagship for OSS and that runs thousands of important sites -- google.com, realaudio.com etc. -- not just my home box here......

What will the next ten years bring? Interesting question.

One teaser: Different architecture of computer/OS for databases? Think about it......