Re: Oracle vs MS SQL Server

In article <01bdcfe3$85396960$0c17b3d1_at_lhodgkiss>, bjohnsto_usa_net <bjohnsto_usa_net_at_dejanews.com> wrote:
>Oracle forces you to do more work up front thinking about how big your
>tables are going to get. SQL server lets you defer these kind of
>decisions. Oracle seems to be optimised to be more labor intensive.

I don't think we disagree here, but I think you've chosen an unfortunate turn of phrase. Oracle can be quite labor intensive, but it's not because it's been "optimised" to be that way, it's because Oracle has always presumed that over the vast number of platforms and sizes of installations that they support, it would be more useful for DBAs to be able to control more aspects of database administration and leave less up to various forms of optimization.

But the key here is a little knowledge. By understanding your data (by which I mean knowing what is roughly going on in the database one is purporting to manage), it is possible to set up a wide variety of defaults in Oracle that will allow your developers to be just as sloppy as they would be in MS SQL-Server. The DBA can then take the road of fixing problems as they arise. Personally, I think it's a bit better to train developers so that they have some chance of creating the applications right to begin with, but in this "real world" of ours, that can be a tall order.

>SQL
>server is optimised to let you be sloppy, at the cost of some loss of
>control, and efficiency. There are some mechanisms to help you reserve
>space for tables, but you have to go out of your way and they seem less
>natural.
>
>The bundled Microsoft front end tools stand out as so far above the
>competition that the comparison is literally a joke.

I've heard that Microsoft is doing a good job with these tools, but I have no direct experience. The Oracle management tools have always been more command-line oriented, but we've never had much trouble accomplishing what we need to do. And since Oracle tends to be used in places with large operations, many DBAs prefer to purchase high quality management tools from third parties.

>Overall cycles are cheaper than programmers so Microsoft is on a good
>thing.

This may be one place where MS oriented IT staff and the rest of the world disagree. One of my fundamental complaints with MS software is that it seems to have the attitude that brute force memory and code bloat is better than doing things right to begin with. I can't agree that this is a "good thing."

>Except that I have such poor experience with Microsoft support I would
>consider their product very seriously. Since they are quite happy to
>ignore bugs in their products which are impacting a customer for and number
>of years and releases I would avoid them for crucial systems. They have
>such an enormous customer base they can't keep them all happy.

My own experience with MS support is even worse. While there are complaints about Oracle support as well, I must say that my experiences with Oracle support over about 15 years have by and large been very good. They have gone through some rough periods, but in general, they have solved my problems responsibly and within a reasonable time frame.

>None of the major databases fully support ANSI SQL92 fully. Even if they
>did, the standard has enough missing features that people would
>legitimately want to use proprietory extensions anyway. All pretty sad
>really.

Oracle does meet more requirements for the relational model than most competitors (although it is deficient in some peculiar areas). I am always perplexed by the lack of a CREATE DOMAIN in the DDL. To the best of my knowledge, Interbase remains the only product which supports domains, and for better or worse that product does not seem to be moving out of its niche. Someone told me that DB2 now has it, but I don't know if that's true. Oracle supports domains in its Designer 2000 package, but in terms of Oracle's own database, this is essentially a kludge.

All the talk about SQL-Server 7 seems to be assuming that Oracle will be sitting on its hands and not improving its core product. While this is not an impossible outcome, I think it is pretty unlikely. And for political reasons, Microsoft is unlikely to attempt to compete with Oracle in the area of embedding a Java virtual machine in its product. So if cross-platform computing is a concern, Oracle is likely to continue to be an important product; if NT conquers the universe, I suppose its all moot.

-- 
-----------------------
Jack F. Love
Opinions expressed are mine alone, unless you happen to agree