I debated whether to post here -- or to the wish list -- or to the mailing list...but I figure with the server farm you have now, using a little disk space won't hurt!
Now that e-smith has gone the free versus $595.00 subscription route (which I argued against
), it occurs to me that e-smith might proceed to market the product almost as if it is an embedded Linux implemenation, rather than a "cd distribution."
There are similarities. By limiting e-smith's functions, and by limiting suport to a subset of i386 hardware supported by the kernel, the e-smith server has many characteristics in common with proprietary network appliances that run embedded Linux from a rom.
There are other applications for which an "e-smith like" appliance/product could fill a market niche. For example: many small businesses, such as ours, have a need for several graphical workstations to provide basic word processing, e-mail, web access, and perhaps accounting/book keeping.
Maxspeed targetted this market with a hardware Xterminal appliance that can connect to a Linux "XDM server," Citrix server, Win 2000 terminal server, etc., over ethernet. At $500.00 U.S., plus monitor, it is more expensive than a discount house PC -- at least here in the USA.
Maxspeed's claim to fame is that by controlling the environment in which Linux runs, they eliminate the difficulties encountered by newbies in configuring both X and ethernet to create Xterminals from PC's.
The main problem with the automated gui installations provided by Caldera, Mandrake, Redhat, et al, is the diversity of graphics cards they encounter and attempt to support. For many newbies like me, simply discovering that a PC could be configured as an Xterminal, was an accident. Few people write about it, and those who do, engage in obfuscation -- making a simple task difficult.
Since e-smith is derived from RedHat, I propose a new product for people looking to expand/replace/create their network computing environment, but who do not necessarily require additional (or any) Windows or Mac machines.
The new product would consist of two parts, and two CDs: The first: a file/print/application/XDM server (but not necessarily running a local Xserver). Any machine capable of running the existing e-smith server/gateway product can do this, but e-smith could publish suggested hardware requirements for good performance with any given number of clients. For example, up to 6 clients: PCI, 586 266 MHZ, 96 MB ram, IDE or SCSCI; 7 to 15 clients, 686, 450 MHZ, 196 MB ram, and so on...
The server CD would install a "user friendly" environment like KDE and some commonly needed applications -- a free word processor like Abiword -- Netscape Communicator for the web and e-smail -- and satisfy dependencies for a handful of other non free applications -- perhaps Applixware, Star Office, Siag Office suites; Appgen Power Windows, provenacct, bestaccounting -- to allow painless installation of a Linux book keeping solution. User accounts might be added via the control console. Printing would set up the same as the concurrent gateway product.
The second CD would provide the Xterminal/workstation installation. Ideally, I'd like to see an NFS install, with the CD mounted on the server, but...
By strictly limiting the supported hardware to specific video cards, multisync monitors, PS/2 mice, IDE drives of say, 300 MB or more, in addition to the restricted list of NICs, a modified RedHat gui install utility would probe the machine, create /boot, /root and swap partitions, install a base system, TCP/IP networking, and a Xserver -- without any other applications and NO window manager. The installation would prompt for a root password, the address assigned to the "XDM server," and then infer the appropriate address for itself after asking if this is "Xterminal #1, #2, #3," etc. Finally, the installation would automatically configure inittab to boot to the multiuser, character cell, networking run level (I guess that's still runlevel 3 in RedHat), and insert a line like:
/usr/X11R6/bin/X -query xdmserver.our.office
in rc.local....
And voila! Server centric, homogenous, network computing, with little difficulty. The perfect compliment to the gateway/router/e-mail server/web server product...
Today, no company can utilize a 100% e-smith solution, unless the only thing they want to do is set up a web server, and a minimal one at that, using vi to create html pages as root.
There is competition out there. With a little Linux knowledge, a 100% Linux solution can be implemented, slowly, and laboriously...but not easily. Storm Linux is attempting this by working on a highly automated workstation/server distro as well as a firewall version, but have ignored, to a large degree, the benefits of a server centric homogeneous environment. Maxspeed is marketing a proprietary hardware Xterminal with embedded Linux, but that still requires extensive configuration of a server.
By arming your customers with a VERY specific hardware shopping list, the task of developing this implementation should be simplified.
The "danger," I see, is that while e-smith seeks to provide an easy and specific Linux solution to Windows users (and does that very nicely), it also gives Windows users a first "taste" of the benefits of Linux and opens their eyes to additional uses.
In order to avail themselves of these additional capabilities, users must turn to "traditional" Linux distributions, such as Slackware, SuSE and Redhat. Once they progress up the learning curve, I believe many will begin to use their new found skills to configure their own firewalls, samba servers, mail servers and print servers. The pattern I predict for many users is, e-smith for a first implementation of Linux, then a gradual migration to a mainstream distro, or a combination of a mainstream distro and various embedded Linux appliances -- ultimately leading to no e-smith at their business.
By taking the offense now, e-smith can provide a family of Linux/e-smith products that make it possible for inexperienced users to implement 100% Linux solutions -- and carve a niche that sits squarely between the mainstream Linux distros and the embedded Linux appliances.
Combining much of the functionality of mainstream distros with the ease-of-use provided by embedded network appliances, and by making a 100% e-smith computing solution possible for many companies -- e-smith can prevent the migration of it's user base to other Linux distributions. The benefit to users would be:
1. Less expensive than Windows, and as easy, if not easier, to set up.
2. More expensive than mainstream Linux distros, but far easier to configure and administer, thus providing lower "total cost of ownership," for companies without a Linux/Unix guru/administrator already on staff.
3. Similar to embedded Linux, in terms of ease of use, but permitting the use of commodity hardware.
Sorry for the excessively long post, but for some reason, I feel strongly that the long term success of e-smith (and companies like e-smith) has implications for the growth and success of my own company. I like the e-smith approach to business, and I'd like to see the company grow and prosper, for entirely selfish reasons...And long winded posts give me something to do while I download Star Office 5.2
Later,
Colin Mattoon
2 Replies
2953 Views