Database Business Problems at Oracle?

abb_road writes "Wall Street responded to yesterday's report of a 42% rise in profits by pushing Oracle's stock down. Despite a 77% increase in applications business, investors are worried that Oracle's core database business remains comparatively stagnant. Though Ellison claims that the DB business will grow in double digits over the next few years, it seems that more companies are switching to open source rather than paying Oracle $40,000 a processor."

Re:mysql

[networkBoy]

Yes.
MySQL will fall flat on its face far sooner than Oracle will. If your DB is tens to hundreds of terabytes, with gig and larger entries (think VLSI design here) then MySQL will not hold up (well). That said there are other OSS db's that will hold up better, though they are slower.
-nB

Re:Postgresql vs. Oracle flame-war.... GO!

[Anthony Boyd]

I found a fairly good review of Oracle, Postgres, and MySQL. All sorta recent versions, too. You can read it here:

http://www.suite101.com/article.cfm/oracle/115560 [suite101.com]

However, it doesn't really get into nitty gritty. Nice primer, though.

-Tony

No one really pays $40k/processor

[darylb]

$40k per processor is "list price". In reality, there are other options, such as Kunta Kinte [slashdot.org] points out.

Further, the kinds of companies that have huge investments in data centers (Oracle's primary target) negotiate volume contracts with Oracle. These contracts push that $40k sticker price way way way down. (Previous employer paid under $20k for a typical Oracle server license, unlimited users, no time limits.)

Considering that these companies really need their data, and have hundreds of applications (not all of them even cataloged) already written to use Oracle, this money is just basic business expense.

Re:Postgresql vs. Oracle flame-war.... GO!

[Anthony Boyd]

Dang! 5 minutes after I post the link, the server goes down! It's an NT server... figures!

Here's Google's cache:

http://72.14.203.104/search?q=cache:3Z3Pzf07oboJ:w ww.suite101.com/article.cfm/oracle/115560+&hl=en&g l=us&ct=clnk&cd=1&client=firefox-a [72.14.203.104]

-Tony

Re:I don't know about open source...

[badmammajamma]

The thing to keep in mind is not what your DB needs are now, but what they will be 2 - 5 years from now. Will SQL Server still be your best choice? Moving from one database to another can be very painful and extremely expensive.

And of course, there's always the "nobody gets fired for picking Oracle" argument. :)

Prices listed wrong - some clarification

[Belgarath52]

Look - I'm as much an open source fan as anyone, but the fact is that the $40,000 figure is misleading. Oracle's so-called Standard Edition One is basically the full thing - it just can't do clustering, and can't do more than two processors.

I'm sure someone will point out another nitpick that it can't do, but the practical fact is that you can buy Standard Edition One for $5000/processor and get a fully functional database.

For the price-aware, you can even buy a 1, 2, or 3 year license for something like $2-3K.

And, no, Oracle isn't paying me to shill for them. I just work for a company that uses Oracle, and I hate to see the "Oracle costs $40,000" meme repeated here.

off-the-bat comparison

[^Z]

Oracle's strong points over PG:
- speed
- mutli-way replication
- multi-node clusters
- advanced SQL (cubes, trees, etc)
- finer details of physical data layout (cluster tables, partitioned tables, etc)
- stability (unless you use the bleeding edge, which is brittle, alas)

PG's strong points Oracle:
- price :) (probably including support)
- relative simplicity and lower resource consumption
- easier administration
- good compatibility with Oracle's SQL ;) (easier migration)
- source availability

Also, PG is perceived as less stable than Oracle, and even less than MySQL. It will take time to dispel this (if untrue).

Re:I don't know about open source...

[Anonymous Coward]

SqlServer is a dog from a performance standpoint. If I was given the choice of Oracle versus SqlServer, I would pay the extra $35K. Seriously....

I spent six weeks optimizing a bunch of SQL stuff for processing once. We supported both MSSQL and Oracle. The MSSQL I finished in about three days. The Oracle work took the rest of the time. The MSSQL processing ran in 1/4 of the time the Oracle did on the same hardware and equivalent schema. MSSQL's SQL optimizer is *so* much better than Oracle's it isn't even funny. You can get high performance out of Oracle but you have to really pay attention to what you are doing. For example, if you have an index on (x,y), if you query WHERE y= and x=, it won't use the index. MSSQL had no problem rearranging the WHERE clause, for example. Additionally, with Oracle to get the best performance, you have to deal with physical disk layout and write your SQL accordingly. You can do that on MSSQL but you can still get 90% or better of the best just by letting the thing do it all for you.

Of course, if I was given any choice, I would probably pick MySQL. MySQL is light, fast, forgiving, and pretty scalable.

Yeah, we use MySQL a bit now and it's fine if you just want glorified text files by default. If you want transactions and the like, you have to do non-default things, which is OK I guess. However, most people who use MySQL don't think transactions are ever needed (mostly because they don't understand what they are and why they are needed). Plus, there are lots of places where your "forgiving" statement seems to mean that MySQL can give back data to you that you didn't think it would (as in... when is NULL not a NULL? see MySQL Gotchas [sql-info.de] for some interesting behaviors that are... fast and forgiving, I guess).