Re: Re : Losing out to SQL Server

  • From: Niall Litchfield <niall.litchfield@xxxxxxxxx>
  • To: David Barbour <david.barbour1@xxxxxxxxx>
  • Date: Fri, 22 Feb 2013 16:48:27 +0000

On Feb 17, 2013 3:34 PM, "David Barbour" <david.barbour1@xxxxxxxxx> wrote:
>Quoting Dan Morgan
>
> " . I'm not
> sure exactly why, but SQL Server has always had trouble with corruption of
> these link-list pointers. Run DBCC to check (and fix) problems with these
> pointers. No comparable issue exists within Oracle."
I like Dan but I do have a couple of problems with this statement, the
first being that I've never seen this, I'd be interested if anyone else
has. The second though is that I have seen a really nasty bug (ORA-00600
kdsgrp1 where Oracle can't find a row continuation because the pointer from
the row header points to the wrong block ) now I'll freely admit this isn't
a linked list issue, but it is a corrupt pointer issue, and of course
theres no dbcc to fix it.
> Additionally, Oracle is still way ahead of SQLServer when it comes to
> Scalability, Load Balancing, High Availability, and Recoverability.
>
> Yes, there are plenty of 'workarounds' in SQLServer to produce
> functionality that approaches Oracle in these areas (see MSDN or just
> Google) but many involve a degree of cost or management
expertise/attention
> that makes them somewhat unsatisfactory.

I'd argue that any proper scalability or availability solution requires
significant expertise and management

As an example, remembering that a
> SQLServer 'Database' equates to a tablespace in Oracle,

A tablespace in Oracle does *not* equate to a database in MSSQL (well see
how pluggable databases work in 12c mind I bet they turn out to be an
extension of transportable tablespaces), nor does an MSSQL database equate
to a schema in Oracle. A database in MSSQL is, just as in Oracle a set of
files managed by an rdbms instance, it can contain multiple schemas,
multiple users, multiple files and filegroups and so on.

I'm not particularly arguing for or against Oracle for the requirements
here, but I do get annoyed when arguments are presented as fact when they
aren't.


--
//www.freelists.org/webpage/oracle-l


Other related posts: