Re: Informix vs. Sybase vs. Oracle vs. (gasp) MS SQL Server

Jim Smith wrote: > I wouldn't go as far as essential but it certainly helps. I would guess that > most applications do not have serious > concurrency problems caused by locking, but for those that do the fact that > a single row update locks one row as against many > rows must help. It would help only if the application wasn't developed with a clear understanding of the issues and processes involved. Locking should be fast. Only lock when you need to within the scope of the real transaction. This should be fast. And this type of lock would be transitory. > I can't speak about Informix, but the Sybase locking model where readers block > writers is probably more of a constraint > (compared to Oracle ) than the page/row difference. Only when the readers hold their lock for longer than it takes to do the actual read. The concept of holding a read lock until a potential insert/update transaction is specious. > > Since when? The level of locking should be transparent to > > well written applications. It only becomes an issue with > > poorly written ones. > > This is the techie answer which frequently comes up with Sybase. Maybe, but its the correct one. > An alternative reading is that you have to code round > limitations. You have to do that with any system of course > (except for my new application development tool Magic(tm)), but it > seems to come up more often with Sybase. Being forced to think of a better design pays off better in the long run. I suspect that any application developed on Sybase first would be inherently more portable to other backends than vice-versa. > Either way, if you always have to employ top-flight programmers, then > that increases the cost of ownership. I don't buy that argument. Having at least one superman in a bunch of turkeys may cost slightly more up front, but it works out cheeper in the long run. Think if you had no top-flighters. How much longer would your project take with just the turkeys trying to get the work done? What would you end up with? How many problems would you have to resolve after deployment because none of the turkeys had the nouce and experience to know better. I've seen it happen time and time again over more years than I'd care to remember. How much more does it cost to have to rewrite a badly designed and written app than to do it properly the first time? How many projects fail? This has been well documented by researchers. You NEED people who know what their doing. -am