Open source strategy - a sword that cuts both ways? (IT-Director)
This is bad news for open source enthusiasts. Fans of the open source movement would, not unnaturally, like to see open source products adopted as strategic. But who are the vendors that are most likely to be accepted as strategic partners by users? If you think about databases it is going to be IBM (Cloudscape) and CA (Ingres) at the top of the list."
Posted Apr 4, 2005 19:08 UTC (Mon)
by szoth (guest, #14825)
[Link]
So the guy wants me to believe that if MySQL AB goes away so will MySQL? I don't even know who all supports PostgreSQL these days, but it seems to be gaining momentum despite the troubles at Great Bridge.
". . . the more it plays back into the hands of the traditional behemoths."
I suppose he means the way Linux is playing into the hands of Microsoft and Sun . . . and DEC.
Posted Apr 4, 2005 19:11 UTC (Mon)
by iabervon (subscriber, #722)
[Link] (2 responses)
Posted Apr 5, 2005 0:08 UTC (Tue)
by jd (guest, #26381)
[Link] (1 responses)
I believe it was Harold Wilson, former British Prime Minister, who claimed that England would never need more than three computers to serve all of its needs. (Which is why there are exactly three Regional Computer Centers in Britain to this day. They have more than one computer, these days, but the whole infrastructure design was based on that belief.)
The problem was not that he was wrong. The problem was that there wasn't anything to be right or wrong on. The question he was trying to answer was based on flawed assumptions and therefore really didn't exist to have an answer.
The same is true here. There is an assumption that databases solve one specific class of problem for one specific class of data. Now, obviously, if you define databases that narrowly, there is going to be an upper limit on the number of different database products that can coexist. However, let us look at that assumption. Does it hold up?
Payroll (which is inherently heirarchical in structure) would seem to be very different from GIS (which is inherently grid data) would seem to be very different from conceptual data (which is inherently object oriented) would seem to be very different from ordered data (which is often sequential or sequential indexed) would seem to be very different from fuzzy data (no fixed relationships between any two elements) would seem to be very different from sparse matrices (grid-like, with too many holes to actually treat as a grid effectively).
On top of that, the optimal way of processing small data sets is very different from handling mid-sized or large data sets and vice versa.
On top of that, the optimal way to implement an algorithm is dependent on the hardware used, which is typically market-specific. Home user x86 PCs are not wired the same way as high-performance x86 servers, even though they nominally run identical code and have 100% interchangable components.
On top of that, have you ever known systems to have 100% uniform loading? Typically, loading is going to vary in a very application-specific and context-specific way.
On top of that, the more of these needs you meet, the larger the footprint of the product, but that is in entirely the opposite direction to what you need for, say, hand-held or embedded systems, where Small Is Good.
It is therefore absolutely impossible for one, or even a small handful, of products to meet all these requirements, and the better a system meets some, the worse it will meet others. Furthermore, there aren't the database products on the market (open source and closed source combined) which cover the full spectrum. I am surprised there aren't MORE solutions being added to the Open Source world, not less.
Posted Apr 5, 2005 15:41 UTC (Tue)
by GreyWizard (guest, #1026)
[Link]
Why can't a single database support tuning that would allow it to handle payroll, GIS, conceptual, ordered, fuzzy and sparse matrix data? Why can't it handle small, medium and large data sets differently? Why can't it effectively adapt to load and have a modular structure suitable for everything from servers to hand held computers?
I don't say that if one database was reasonably good at everything there would be no room for others. For example, some say FreeBSD is better than Linux for routing and Plan9 has innovative features such as the plumber that have no equivalent elsewhere. I see no reason why things wouldn't be analagous for databases if one project had the kind of momentum Linux does. But that doesn't mean that one project can't meet all of the major database requirements as a matter of principle.
Posted Apr 4, 2005 19:59 UTC (Mon)
by jd (guest, #26381)
[Link]
MySQL is still one of the fastest SQL databases for simple databases, though there are now some nice SQL-based libraries out there for those not wanting to run their own engine. For dynamic web content, MySQL is generally the best option and I don't see that changing any time soon. For dynamic applications, you don't need a heavy-duty engine and typically don't have vast amounts of data. The worst that'll happen to MySQL is that it'll become a database for specialist functions.
PostgreSQL is probably the best general-purpose mid-range engine out there. It's not as scalable as Ingres, not quite as fast as MySQL, but if you're not looking for these ultra-specialised sorts of data sets, then why would you want a database engine that was ultra-specialised? That makes no sense. PostgreSQL is ideal for what it does, and I don't see any commercial venture coming close for as long as it remains an actively and aggressively maintained system.
The problem with Techno-Prophets is that they assume that the niche they happen to like is the entire world. It isn't. There are plenty of other niches out there, some bigger, some smaller. There is also plenty of demand for non-specialised systems that'll do everything but some specialised task better than ANY of the specialised systems.
Personally, I'd like to see more database applications exploit the strengths of multiple engines, rather than be hampered by the weaknesses of any. In such a vision, the Open Source strategy is guaranteed a slice of the market that will lie between 1/3 and 1/2 of what is there. That's fine. I'm happy to leave the other 1/2 to commercial operators - they have to eat too.
Posted Apr 4, 2005 20:37 UTC (Mon)
by khim (subscriber, #9252)
[Link]
Gosh. Have they learn anything over time ? Open-source is different then closed-source. It does not really matter how big is your company - what does matter is how big is your community. Even if company will go bankrupt while community is thriving you can be sure product will continue and it will be well-supported (think Eazel and Nautilus). And if company does well but community is not there then... oops (think IBM and NGPT). And right now time works for MySQL and PostgreSQL: there are more users for MySQL and PostgreSQL then for all "traditional databases" combined. In Open Source world money is just money: tool to motivate developers. Not even the most important one. It's not "the god" like in "traditional behemoths" world.
Posted Apr 4, 2005 20:57 UTC (Mon)
by mchristensen (guest, #4955)
[Link]
Posted Apr 4, 2005 22:09 UTC (Mon)
by lutchann (subscriber, #8872)
[Link]
Isn't this much better than the "Open Source is for amateurs" and "Open Source will get you sued" lines that were coming out of these places just a couple years ago?
Posted Apr 6, 2005 18:42 UTC (Wed)
by przemek (guest, #26930)
[Link]
Of course it is a good practice to have an eye for vendor lock-in with any technology; one avoids it by selecting suppliers that don't play the lock-in game, and by paying attention to implementation details, refusing or at least controlling the use of unique, non-standard features. Open Source is just consistently making it easier to avoid the lock-in.
Why does Open Source have less lock-in? The source availability means that truly useful features propagate to all competing systems; also, the customer expectations are against such practices. It is therefore easier to deploy systems that are not critically dependent on any particular component. MySQL not good enough? fine, I will run my SQL queries against Postgress. Can people paint themselves in a corner? Sure, but it is not because they had no choice.
". . . at some point in the future the market will consolidate and a number of these products will disappear."Wishful thinking
This article makes the odd assumption that the maintenence of an open-source project depends on the money available to a company working on it. But IBM in particular is known for discarding anything that they don't need to do themselves. So I suspect that IBM will decide at some point that HSQLDB satisfies the needs of their customers and ditch Cloudscape, like they've done with Jikes, their hard drive division, and their desktop division.Open source strategy - a sword that cuts both ways? (IT-Director)
It also assumes the total money available is fixed, so that more products equals less potential marketshare and therefore less money. The market's growth is largely independent of the number of products. It might grow faster, equally fast, slower, not at all, or in reverse. Sometimes, it acts as if it is doing all of the above.
Open source strategy - a sword that cuts both ways? (IT-Director)
While I generally agree with your thesis that there is plenty of room in the database market I am not convinced that different workloads and footprint requirements necessarily requires a completely separate database project. Most of the arguments you've made could also be made about operating system kernels too. Surely no one kernel could most effectively handle both interactive and batch tasks, let alone run efficiently on a dozen differnet CPU architectures or scale from embedded systems to multiprocessor supercomputers. Except of course that Linux does. While it still isn't a good fit for some niches (realtime comes to mind) it is gradually improving even for those. Open source strategy - a sword that cuts both ways? (IT-Director)
Ah, this must be why Ingres is now Open Source. True, DB/2, Informix and Sybase only have Linux ports, rather than being opened themselves, but it would not be impossible for that to change on Informix, as IBM are trying to get supported clients to move to DB/2.
Open source strategy - a sword that cuts both ways? (IT-Director)
Open source strategy - a sword that cuts both ways? (IT-Director)
Popularity and success with early adopters may not be an indicator of future success with the pragmatic folks who are still playing wait-and-see with FOSS. But that does not mean that it is insignificant, and the old "big vendor wins" rule will be the only factor in determining which free databases will make it in the enterprise.
Prognostications like those in this article are easy to make, but hard to make correctly.
--Mark
Open source strategy - a sword that cuts both ways? (IT-Director)
Keep in mind that Philip Howard is the Research Directory (Technology) at Bloor Research, who sells IT industry analyses for lots of $$$. He writes free (as in beer) articles such as these as nothing more than marketing to try and convince IT managers that his high-$$$ research reports are unbiased, well-thought-out studies suitable for long-term planning. Here, he's putting on a "Open Source is good as long as you have a support contract from a respected company" hat, which seems to be the standard line among conservative industry analysts these days and probably goes over well with managers still clinging to the single-vendor mentality.Open source strategy - a sword that cuts both ways? (IT-Director)
The article has valid points about the different level of expectations for technology of strategic value, but ther conclusions are flawed because the author misses one of the fundamental principles of the Open Source methodology: conformance with Open Standards! In a standards-conforming environment, you can buy an Amana refrigerator and when they go out of business or it breaks, you replace it with Frigidaire---unplug the old one, plug in the new one. It'll cost in capital and labor, but it's not a tragedy. Open source strategy - a sword that cuts both ways? (IT-Director)