Re: Help to Promote Linux! - Economics....

Greg (greg@cyberramp.net) wrote: : Just fer sport (and I really don't have a preferred side here...), how : do you all see the economics of Informix on Linux working? All of the : Linuxes I'm aware of (and I'm nobody's expert here either...) are : inherently uniprocessor machines, mostly PC's. So basically, we're : talking the lowest end, most price sensitive segment of the market. [Snip] Well just for sports then. I'll step over your financial statements with regards to Informix. I just don't know about that. Indeed Linux works in the low end market since it is free. (It has been testified many times before in this group that that is the place informix came from. And that they want to play with the big boys now. A quit ridiculous idea only caused by the fact that informix is perceived wrong in the market place. Which, again testified many times in this group, is caused by marketing problems. I would say that Informix has been a bit unlucky in the past) But look at the effort going on the way to make Linux 64Bit, rewritting the file system in the 2.1 development, the SMP development. It is a low end market to ACQUIRE Linux, But Linux could work in a high end market. (I think that I could port a system to a fast Linux intel machine which bills something like 40 Million Dollar per year. Well porting wouldn't be a problem getting it running fast enough might or might not pose a problem) But alas we call Linux low end market (to acquire) and there for it justifies for Informix to port SE, Esql/C, Esql/Cobol, Esql/Ada,c4gl,r4gl and Cisam. (Compilers for Cobol Ada and Fortran are available under Linux Commercial and non commercial and since most preprocessors are layered on top of Esqlc it shouldn't be much of problem to make them available). All of which are simple to port to Linux and would not require much support effort. (Since Sco ports already work on Linux it must be simple and could not require a big investment from informix. There for the business case argument from Informix seems moot in this respect. I have been involved in discussion of this sort and heard the arguments but lets keep it to common sense: A port all ready exists and porting, of the 'simple products', would not take much effort. Testing should be fully automated by now and I'm pretty sure that someone would volunteer a Linux 2.0.32 system to do the testing on. Probably those products will get a better test then any product ever before in Informix when available for Linux. Make that a steady release for a year and release new port one year later; After IBM, HP, SUN and SCO ports. Keep the release to once a year in order to cut down on costs. ) The pricing model of most major database vendors is such that it doesn't work for Linux. SE (1 Cd) cost 1500 US dollars over here. Offset that to 15 US dollars for a Linux (5CD's !) with unlimited and TIMELY upgrades and patches for AS LONG AS volunteers are available. I don't see a real market for Online (OWS and the works) on Linux. Because if you are going to use databases that big then you probably want to use mirroring and a volume manager. And having Big Money in a database justify to invest more money to keep it save. And beside that Se gives a good performance anyway without to much trouble in seting up for small to medium sized databases (What ever that may be). (Especially since the ext2 file system which, I here say :-) , is one of the fastest in the field. ); If you want to develop on top of online then a network connection would do any way. You can then develop on a low cost machine against a high cost (and hopefully) well performing machine. There for Informix must also port the networking stuff. But off course they can do without that since other vendors already sell middle ware for Linux and since, I assume now, the porting of networking stuff is more involved leave that to the middle ware parties. (Isn't there a free ODBC driver project for UNIX's which might address this to, DBD:perl might address this issue also). There is a way where informix could benefit from the Linux model, Since 4gl is called a end of life product hand over the 4gl code under GPL to the Unix/Internet community and let them work on it. This way informix will get a much better 4gl with timely upgrades and new developments. And Informix doesn't have to support it anymore. Which probably saves them a lot of money. Looking at the results from that community then I wouldn't be suprised when they will make it work on many databases :). And probably make it work in GUI environments to. On the other hand if someone in the community set his/hers mind to it the community probably could build the suite (CISAm, SE, ESQL/C and 4gl) in a year or 2. As to support. Lets be short on this I do not believe that there is one vendor who provide better support then the Linux community. Since that is free and better, no mandatory charge for support unless something substantial is delivered for the money payed. (Probably most vendors don't have a track record for providing something good here !with some exception to the rule off course) My conclusion. Porting of Cisam, SE, ESQL/xx 4gl (rds & compiled) is feasible and not labor intensive. Business case argument is moot since existing ports already work. Online (OWS and the works) not feasible for informix to do on Linux. The prices have to come down a lot. SE 200 US Dollar per user. 4gl, Esql/xxx development one time investment (400US Dollar) and no runtime pricing OR development free and runtime pricing (50 US dollar per user) CISAM Don't know but somethine along the line of 4gl. My one cent for sports -- Rene de Zwart