Breaking Down the Linux MarketsAug 15, 2005, 23:30 (12 Talkback[s])
(Other stories by Brandioch Conner)
By Brandioch Conner
You've seen them before. The titles vary but the material all boils down to two topics: "Linux sucks because I had problems getting to work right" and "Linux sucks because it doesn't have have 'x' so it will never be accepted." The former usually has the person trying to install Linux on a new laptop while the latter is usually about different installers or Microsoft Office. They may be right, but I don't think so.
These comments usually focus on what I will call the "home desktop segment" or the "gamer segment" and completely miss the characteristics of those segments. Aunt Agnes (in the home desktop segment) is not going to change her OS on a whim. This is also the segment that collects all the spyware and gives rise to the hordes of spam zombies.
Now, a tiny bit of history. Microsoft did not start off with a server OS. They provided the OS for IBM's PC. They got into the server room via the desktop and the desktop is still their bastion. Even today, most people become familiar with Microsoft's desktop products long before they work with any of Microsoft's server products.
Linux is taking the opposite approach. Linux is starting in the server room and slowly moving into the desktop segments. Emphasis on the "slowly" right now. It takes a lot of effort to move into a monopoly and Microsoft still has their desktop monopoly (yes they do, the court said they did, and the appeals court did not contradict that). In order to understand this better, you have to break down the various characteristics of the desktop and group them into identifiable segments. I've broken them down into four segments (server, corp/gov office, home desktop, and gamer/power user) based upon the characteristics I will identify (including their incentives to run Linux).
Dedicated admin: Someone is responsible for that server. Someone is called when it crashes. Someone installed the OS and app(s). Even if the admin is nothing more than a consultant/contractor who is brought in when something goes horribly wrong.
Many users per box: When a server crashes, multiple people get annoyed, all at the same time. Then someone contacts the admin or power cycles it themselves (very scary situation there).
Limited hardware selection: SCSI cards and NIC's and so forth are the basis of the server. You will not be plugging your new digital camera into the server. You will not try to put your server in "sleep" or "hibernate" mode. You do not sync your PDA by plugging it into your server.
Admin installs the OS and apps: The OEM might include a "quickstart" CD or something, but the server usually arrives without any OS or apps on it. The admin has to choose the OS, how it will be configured and so on and what apps will be installed and how they will be configured.
Functionality: If it doesn't do what you want, why would the admin be running it? Lose that and get something that does do what you want (provided such functionality actually exists).
Stability: Remember the "Many users per box" characteristic? Stable servers mean less stress for admin (and power cycling the box is a good way to test the MTBF rating).
Price: I list this last because if it doesn't have the functionality you need and the stability then even "free" (no cost) is too expensive. The cost of the licenses is only a portion of the "total" cost of the system, but it is a portion and if it has the functionality and stability, then you should see "economy of scale" kick in.
You should see why Linux would make its initial gains in this segment. The functionality of the LAMP stack and Linux's rock solid stability (and low price) put it at the Internet edge of many companies. From there, it took on email, database, authentication and other functionality. Now there is no function that you cannot find on a Linux server. I said "function," not "application." Running Active Directory on a Linux server will present problems, but Novell will sell you eDirectory for Linux. So Linux can provide all the functionality requirements for the server market and Linux's market share in this segment keeps increasing.
Corp/Gov Office Segment
Corp/Gov Office Characteristics
Dedicated admin: While there isn't the same ratio of admins to workstations as there is in the server segment, there are specially trained (it is hoped) technicians who will fix any problems with the workstations.
One user per box: As opposed to the server situation, there is only one user per workstation in this segment. Because of this, the "stability" aspect is not as important (although it is still a factor).
Standardized desktop box: This is another one of those "economy of scale" things. It is far easier to admin 1,000 workstations if they are all the same than it is to admin ten if they are all different (personal experience). Corporations and government agencies tend to buy/lease in large numbers.
Admin installs the OS and apps: The workstation is configured to a standard that has been decided upon by those who (supposedly) know best how to make it run the apps that are installed. The person actually using the box will not be installing any apps or upgrades or changing the OS.
Training for users: Since most users will probably not know how to use the specific apps (other than mine sweeper and maybe MS Office), they will need some type of training by the company. This training should not be much different between a Windows box and a Linux box.
Corp/Gov Office Incentives
Functionality: Same as in the server segment. If it doesn't run what you want, why not switch to something that does? This isn't so much a "not present" in Windows as a "comparable functionality exists in both systems." On the corp/gov desktop, there isn't much functionality that isn't already available for Windows. This is usually a limiting factor for Linux on the desktop in this segment because the systems have evolved over time to their present state. That means a lot of cruft and baggage that has to be handled in some fashion. Just the presence of those "legacy" apps will not rule out a Linux desktop, but migrating that functionality will add to the migration costs.
Stability: Not as important as in a server, but it has to have a certain degree of stability (no hourly crashes). As long as they don't crash between their regular cycles (nightly or weekly), they are stable "enough."
Price: Corporations and governments are usually looking for ways to cut costs. If one system has the functionality and stability necessary, but costs less, then it's all about the pain of migration.
Potentially ideological: This can be the most important one for non-US entities. Licensing Windows from Microsoft means sending money to Bill in Redmond. Even if some of it stays in the local economy, simple math will show that more of it would stay if they went with Linux (unless they hire Microsoft programmers from Redmond to write the Linux apps). Even if the other incentives are not sufficient to justify a Linux migration, the ideology might be enough.
I believe this will be the segment that Linux takes over after the server segment. The first reason is that it offers huge license savings, but not in the obvious way. Even if the cost per desktop is the same as for Windows, the money will be paid to local developers (local for that government) who will pay taxes on it and pay their mortgages with it and so forth. So the government will be getting back a larger portion of it in taxes than they would if it just went to Redmond.
The second reason is that the lessons learned and technology gained from each departmental migration can be applied to the next departmental migration. This means that the second department to migrate will already know what workstations work well with Linux and where to go for Linux ports of their apps (any that weren't already ported).
The final reason (and to me, the most important) is for 100% open formats for their documents. So it won't matter if the government uses one word processor and one of their vendors uses a different one, as long as they both have the filters to handle that format.
Home Desktop Segment
Home Desktop Characteristics
Limited support from vendor for limited time: The OEM may provide support, or may not. Certain companies are notorious for trying to find any way to deny any responsibility for providing support. If Windows came pre-installed then Microsoft will not provide "free" support. This was a great move on Microsoft's part and it saves them a TON of money in support calls. Linux vendors will not> be able to duplicate that trick.
One user per box: See above.
No standardization of hardware: Almost everything out there will be found in this segment. If someone made it and sold it, someone in this segment bought it. If you grab 100 home machines at random, you can have 100 different video cards, sound cards, NICs, and so forth. This is the exact opposite of "economy of scale."
OS was pre-installed by OEM: The hard work was already done. The drivers were found and (hopefully) tested before the user could even order the box. The OEM can provide support because the OEM knows how it was installed and what the components were. Good luck getting that with an after-market OS.
Microsoft monopoly: This means that those weird components will come with Windows drivers. They may be sucky drivers and they may cause problems (BSOD's), but the devices will appear in Device Manager and it will at least appear that something good is happening.
Multimedia: Video and sound drivers--the current bane of Linux. If the home users cannot watch the cute video clips or hear the funny sounds, they will not be happy with Linux. So the video and sound devices have to be supported and the distribution has to have the codex necessary for playing file.
Home Desktop Incentives
None: If it didn't come pre-installed, it isn't going to dent this market. No matter how easy it is to install or how pretty or how secure or how stable or how... whatever. This segment will not even install anti-virus apps (yes, there are exceptions).
Important Note: If someone is willing to be the dedicated admin for a home desktop user, then none of this applies and the relationship becomes more like the "Corp/Gov" example above. The "admin" chooses the distribution, checks the hardware, installs the OS, configures the apps and shows Mom/Grandma how to use it and is (most importantly) available for any questions and problems.
Gamer/Power User Segment
Gamer/Power User Characteristics
No support: These are the people who will upgrade their video cards, overclock their processors and such. They're on their own for troubleshooting.
One user per box: See above.
Microsoft monopoly: See above.
Multimedia: See above.
Latest hardware/toys/games: This is the biggest problem with this segment. First off, they get the stuff before the Linux developers do so there's really no way for drivers to be included in Linux. Secondly, the stuff they get is version 1.0 stuff and usually hasn't had all the bugs worked out.
Gamer/Power User Incentives
None: This segment is all about the vendors. Did the vendor release a good Linux port of their game? Did the vendor include good Linux drivers for that video card? What is the vendor's motivation for releasing anything for Linux? Usually, the vendor is motivated by sales and profit. The Windows desktop market is still at least 10x larger than Linux's so that at least gives the opportunity for 10x more sales if they focus on Windows.
So, when some journalist complains that Linux sucks because it will not install on his brand new laptop (90% of the components are correctly detected and configured but the wireless card doesn't work and the video card doesn't have hardware acceleration) it is clear where he missed the mark. He is complaining that Linux is not ready for the fourth segment when it is still growing into the second segment.
The same with someone saying that Linux will not succeed until it has a better way for the user to install apps. The first and second segments do not allow the user to install apps and the third segment is blocked by the mountain known as "Not Pre-Installed."
Each of these segments will have "low hanging fruit." That means that there are certain people who will have no problem installing and running Linux as a home desktop even though most corporations have not adopted it in that role yet. And in each segment, certain sites will have a very easy time of migrating while others will have a very difficult and/or expensive task if they choose to migrate. This depends upon how tightly they are tied to specific apps and the cost of migrating their data from those apps.