Re: Informix running on Linux

Nils Myklebust wrote: > > This is greate from Tim, David and all the others. > > If I had 7.x OnLine on Linux and 4GL, Java, JDBC and ODBC (I would > make sure that the two last where available form OpenLink immediately) > that's the products I would use for development. > After a short while of developing with them customers would buy my > products to run production on Linux. And my products are mission > critical as few others. > > There are several reasons why this is interesting for me. > One is the price which is getting out of hand using SCO Unix which is > my alternative. (Sun, IBM and HP are fine for large customers, but > their prices are becoming realy stupid.) I was hoping someone would comment on SCO. I have no recent experience with it, but did work with it a couple of years ago. I am very curious as to other experiences with it, in light of Linux. However, the points to Linux over the other x86 Unixes are: 1. Low cost - it's free 2. High value - more products for it than any other UNIX 3. Better scalability - it runs on virtually ANY PC. These all add up to a total cost of ownership for Linux that is a dream come true for many businesses world-wide--especially in developing countries and/or in towns and cities where many businesses can't ever afford systems like SCO or Solaris. Even if the OS were cheap, other software for it is through the roof in cost. You may recall a few years back SCO was bundling Ingres with one of the first System V releases when they were breaking away from Xenix. It flopped because nobody cared about Ingres. Interesting that Informix has ultimately bought Ingres in Dr. Stonebreaker. > Another is support. I can easily get support on Linux from a lot of > sources. Using SCO I know of 2 commersial organisations here in Norway > that can do it. One isn't very good bussinesswise. The other bind me > to their PC's only - a local brand. The situation doesn't seem to be > all that much better in other countries where I will sell my > applications. This is another good point. > A Linux port would make it easier to sell to a large number of small > customers while I would still be able to sell to larger customers > using other ports. At some time even the large ones may use Linux. > And at some time the smaller ones may make more money, and upgrade the hardware, and retain the OS. This is the opposite of the past, where you upgraded to not only new hardware, but to a new OS as well. > I still think Informix should also evaluate porting SE to Linux. They > have a very large number of SE sites out there (500.000???). I think > they are currently loosing these at a fairly rapid rate, possibly > without knowing about it. Large numbers of them are on old (5 to 10 > years or older) hardware and badly need a change. Many are still on > ver. 2.x (I know of many in Norway alone) and 4.x. Many simply need to > do a few new things in 4GL that the old products can't handle. Others > need connectivity for new tools or simply to MS Windows which the old > engines and/or hardware it runs on can't handle. > Informix have no upgrade path for these users. > ***No upgrade path.*** Linux on PC --> make mo money --> upgrade to Linux on Digital Alpha, retain investment in software, and OS. > The only real option they currently have is to rewrite their > applications for other databases. > Does Informix even begin to understand what this is doing to their > market and marketshare? > > Although David have difficulties running SE, these people don't. Many > of them would have problems with OnLine. However you look at it SE is > very much simpler. 7.x Online does present some problems. However, if the largest base of users were on Linux, tool makers would probably invest in developing cool front end products that folks could use. I'm very impressed with Tcl, and who knows, if I were a tool maker, I'd see a platform that would be worth investing tool development in for the LONG TERM. I wouldn't have that big question mark over my head like I do for NT, or other x86 UNIXes. I'm so glad I didn't but Unixware... where is that one now? BTW, HP has made a big committment to MS, so this is a scary thought as well. > With SE you copy the Informix software to the machine and brand it. > Copy the application programs and database to the machine, set a small > number of environement variables and start running the application. > Nothing more to start it. > Then you set up a cron script that does Unix *and* database backup > nightly and run an "update statistics" script and bcheck/secheck. On > any restart of the machine you make sure bcheck/secheck is allways > executed. > Our users run's for years with this setup with essentially no > management at all - nothing except changing the backup tapes and > checking the backup logs on their printer. > > Many don't even run logged databases. Can you imagin? - They don't > need it. - Nightly backups are fine for many applications. > I know many OnLine DSA shops like this. :-) They're out there. > You can run logged databases with SE, but if/when that need becomes > critical and the amount of logging increases, the time has come to > upgrade to OnLine. > People are claiming they run 10GB databases on SE. That can surely > work well in some cases, but personally I would switch to OnLine > arround 1 GB for most types of applications. > I've used 7.x SE a couple of years ago, and it worked well. I agree that the SE distribution should tag along and be a part of the port. How much can it cost Informix to add it to the list? > The above notwithstanding OnLine should get the highest priority if > Informix need to prioritize. Porting SE should however be a no > brainer. A simple recompile should do (hopefully). Yes! > If there are problems Informix may even considder supporting the > current SCO version on Linux. > Nah, go with the real thing, and port it. Besides, it's not like just porting it to a typical OS. The developers are accessable, whereas on the other hardware platforms, we don't have any idea who to contact. > Informix should also understand that a Linux port would be one great > way to get into the MS Windows based market in a better way. How can > that be? The base market for Informix at the low end have been, and to > a large extent still are the 4GL applications that have been > developed. They require a Unix server. Whatever people try to do, > runing 4GL applications directly on MS Windows or even Windows NT > isn't easy and doesn't work well for these people. With a relatively > low cost Linux port they could get the connectivity they need, keep > their old 4GL applications runing in terminal emulators on Windows and > start developing with newer tools. The new tools could be a US$495 > version of NewEra - at that price many would be interested - Java, > Delphi, VisualBasic, PowerBuilder or any other tool the developer > wanted to use. Yes, this is true. NT is not a cheap way to go for building applications. It seems to grab money out of your pocket almost as fast as it grabs memory. The kernel is basically untunable, wher