Re: Informix running on Linux

Richard Stanford wrote: > > On Tue, 29 Apr 1997, Tim Schaefer wrote: > > > The technology aside, Informix has lost the most important > > pieces of the computer business: DEVELOPERS and CUSTOMERS. So, to > > worry about NT is a red herring. > > [Grin] I guess you're not a stockholder, then. Seriously, Informix's > decisions /have/ to be based, at least partially, on what's profitable. > It could make us all really happy and then go out of business -- that > wouldn't help anyone. Now, personally I don't think that a FreeUNIX > port would do this, but they have to think of the bottom line. > I still don't care... :-) Seriously, yes, I do care. But like in a couple of the other rants I mentioned it's the <how> of the money making that is probably the point. > > Right now, if I were to recommend a UNIX solution for client/server, on > > the small to mid-size end, I'm probably stuck with HP, IBM, or Solaris > > or maybe SCO, depending on the client, and their ability to pay. I know > > damn good and well I'd be laughed at to even suggest NT at many sites as > > a viable SERVER for Informix. If they use NT, SQLServer is probably > > already there. So, that's an uphill battle I don't even want to think about. > > Sure. Bear in mind that SCO is now free for evaluation and personal use > during this debate. > Ok let's back up a minute. It's not just the O/S I'd be selling, or just the data base. This just gets me started, then I can think about other software to put on the machine, and possibly a custom solution. This could also be in the form of a vertical market app, with the ability to customize it. Some of the vertical apps I'm thinking about don't run on SCO, they run on Linux. In the beginning this was the attraction to small systems, as an alternative to bigger, costlier solutions. Many customers <I> would have in mind don't even show up on radar, so to speak. Now that the dust is settling on client server, we can actually revisit some of the past solutions, and see that they really are not that bad--in fact 4GL persists in spite of the supposition that it's dead. :-) DOS succeeded the biggest of all reaching out to the widest possible market. It built consensus on the metric of one-user, and is now a serious contender in the corporate marketplace. Microsoft is biggest of all, off of the market that doesn't show up on radar. Go figure. :-) For a data base vendor to tell me they'd have problems making money off the Linux port what I'm really hearing is that they're in big with some of the hardware vendors, and can't supposedly risk a Linux port. This would appear to "undercut" the other vendors. But if we applied a little rocket science to the marketing as much as has been done on the engine, we'd know that Linux would probably enhance sales of other platforms. Some really creative marketing solutions could come about. A free open market with a free O/S allows for innovation and a real new era. Competition would really get going. > > I'd rather they buy an HP K-series or T-series server, and then connect > > across the pipe with Windows PCs. This method has the highest success rate, > > and is probably dare I say the most popular. But to go to a lesser > > platform for smaller business, the cost is too high. They roll over > > to MS. > > Although a decent x86 based mission critical server runs about the same. > [Sigh] Getting people to believe that, though... but that's another debate. > > > As to Linux's viability in the market: It persists. Anybody who > > knows the small potatos markets will see that if ONE major player > > in the data base market goes to Linux, NT doesn't really stand a > > chance. There's no need for it. The whole point to NT was to > > provide replacement business for UNIX. Plain and simple. But > > because Gates doesn't understand the UNIX market he's missed the > > boat. Now he's using NT to grind on Novell. What a pitiful > > waste. The OS has not progressed, but the overall cost has. > > Wrong. Plain and simple. Say I'm a medium-sized company, trying to > save money but not afraid to spend on commercial products (like an RDBMS) > to get results. I'll probably not set up a mission-critical server on > Linux for one simple reason -- no support. > OK that's for a medium-size system, agreed, but the little systems, these are the point. And a potential migration plan could be planned moving the customer to Linux on DEC-Alpha. You might be surprised to know that many companies are using Linux for mission-critical apps, only they probably won't talk about it as much. I appreciate the concern for support, however,I've got the 2.0.29 kernel, and it just chugs along, not causing me any more problems than Solaris--probably less. I would be more concerned that Informix wouldn't just follow a revision, rather than writing their own. But if they DID release their own Linux version, you know darn good and well, the industry would probably chase after it, make sure their products work with it, and in the long run provide stability. It would of course open up the idea of a "Informix" version, but what's really wrong with that? Why not just go full tilt, produce the port, and let the chips fall where they may? > Say that Informix does a port to Linux, and one of the (many, but minor) > Linux updates comes along. Maybe it fixes a security hole or something. This > update contains a bug that causes intermittent failures in Online. Once > Informix finds this out -- what do they do? Tell users (users, NOT os > developers, mind you) to put in an OS fix? Tell them to contact their > OS vendor (oh, you don't have one? Hrm...). And who takes the blame? > Informix. This is the hardest part of the port -- after the sale support. A good point, however, we see Microsloth succeeding on the same odds. They have produced far more copies of their software than they could ever support. We see a lot of their software released with bugs. People will find ways to overcome direct vendor support. And a low priced software package CAN be supported, it's just a different payment plan. Why not involve us developers out here in the real world, and allow us to support it--those of us that would like to? Keep in mind if you have the source code to the O/S, support problems should be easily identifiable--I'm being over-simplistic,but it's not like Informix would be operating in a vacuum. After all, what is OnLine, but a sophisticated O/S that manages a raw-disk and memory? If we think about it, the cost to produce a complete Informix Hardware Solution should not be so unthinkable. Look at what Oracle is actually attempting with NCs. Informix could do one better with a complete Informix machine using Linux. It would be open, extensible. > > > And the other small potato UNIX players don't understand that > > developers and small businesses are STUCK with NT or overpriced > > x86 UNIXes that perform poorly. So between MS missing UNIX and > > UNIX missing UNIX, there's Linux. It's a great product, and > > right for the time. > > > I refuse to suck up to Informix. I've endorsed their products whole > > heartedly up until New Era, and continue