Re: Sybase vs Oracle - which is better?

Hi,
Sorry, but I simply didn't see your reply until now (sometimes I hate this M$ Outlook Express thingy.

kennedyleigh_at_yahoo.com wrote in message <7341v7$o08$1_at_nnrp1.dejanews.com>...

>> I do find some of your claims to be wrong though ...
>> kennedyleigh_at_yahoo.com wrote in message
 <72rn7e$fpb$1_at_nnrp1.dejanews.com>...
>> Hmm .. I don't know what Sybase Central is - but if it's like Enterprise
>> manager, that's part of the standard client tools for Oracle. Enterprise
>> From the name "Sybase Central" it seems we are talking about some of the
>> same functionality?

=>The enterprise manager (at least on NT) looks a bit disjoint to me. It does
>not support the ability to start and stop oracle processes as far as I can
>tell. Sybase central is more that just a tool for Sybase databases. It has
 a
>plug-in architecture (similar to netscape) which supports managing of All
>sybase products (ie: Adaptive Server Enterprise, Replication server, IQ
>Server, etc.).

Then I'm sorry about your misconception. Every aspect of what you write above is covered by Enterprise Manager. From starting and stopping the processes (called an Instance in Oracle), to plugins that allow you to control and add on 3rd party software too. Ei. Oracle uses this feature to allow you to control Oracle Application Server through the Enterprise manager. Yes, it has modules for replication management too. Yes, the replication can, in the enterprise edition of the Oracle Server, be bi-directional, have multible host tables etc.

In essence I think the products are simular in context.

>Date ? Don't start me off on Mr "I love IBM" Date (Sorry, just a pet hate
 of
>mine !)
>It sounds like Sybase and Oracle's internal implementations are radically
>different. Sybase doesn't use cursors internally in that way.

Hehe ... sorry about that. He just got me started on databases back when computers where humming in large rooms and not on your desktop :) To me, Date defined relational databases - and quite frankly as the SQL language came from IBM, then you kinda agree with me there too :)

A cursor is what you store your SQL command in, in order to communicate with the RDBMS. It's a memory structure that holds information about bind values, pointers to where you are etc. The PL/SQL cursor is a little different, but in essence you are using the same technology, but the PL/SQL engine does that for you.

>You are confused by terminology here. By Server I mean OS Process. The
>Sybase term 'Server' is the same as the Oracle term 'Instance'. What I was
>saying, was One Oracle Instance (ie: OS Process) can only have one Oracle
>Database, where as one Sybase Server (ie: OS Process) can have many
>Databases. Obviously you can run multiple OS processes, but that isn't
 very
>efficient as you are Duplicating System catalog, and other common
resources.

As another participant wrote, I think you're confused about the technology. A database instance is per definition independed. A database can hold many "independed" applications shielded by schemas but contained within the same memory structure and data dictionary. If an enterprise needs 2 application systems that runs independently of eachother, they don't need to implement 2 Oracle Instances to run them. Only if they need independence in backup procedures and emergency recovery operations, splitting them up makes sense (or if the application types are so different that sharing the same memory becomes a problem).

>> Agree! Oracle was one of the last vendors to implement constraints.
 However,
>> the implementation they made seems to me, superior to anything I've seen
>> elsewhere?? (I'll like to swap knowledge on this part).
>How does oracle implement constraints ?

Good question :)
Starting by talking about which constraint types Oracle support, Oracle has primary key, unique, check and foreign key constrants. The primary key and unique constraints ensures a unique set of columns. A check constraint is a boolean expression that has to be true in order to accept inserted or updated data (check constraints only works on A record). A foreignkey constraint insures consistence between tables, making sure that referred to values exist in referred tables.

On top of the constraints you have triggers. Triggers that can validate and manipulate your data in any way you want (well, ALMOST - did anyone say Mutating here??).

What abilities does Sybase have?

>Sybase are an Open Systems company. They believe that Customers should be
>choosing the Techology they use, not a couple of idiots involved in a
 public
>slanging match (and you have to admit, Larry Ellison & bill Gates both say
>some stupid things sometimes).

Totally and utterly agree :) They do both talk stupid from time to time in their popularity contest. On the other hand, every previous philosofer has been marked by his fellow human being during their (too) early predictions ;)

Thanks for the info about Sybase !!

>I get the impression from you statements that Oracle is so different
>architecturally the original issue may not be comparible between the two.

Which original issue? My point is, that both has a place on the market. Depending on the application and scalability needs. What's important - to both of us - is to keep our mind open for input from the "other side" and try to evaluate and see where which is better for what.

• Peter Larsen Senior Oracle Consultant