As an admittedly biased Oracle architect/DBA, I'm inclined to believe that the "Oracle" bottleneck was due to the design/implementation of the Oracle DB, rather than the technology itself.
If, for instance, you were pulling this data out of an Oracle instance designed to be a cheap, historical archive, on minimal hardware and cheap disk, then yeah, it can be very limiting in performance.
If you're pulling it out of a production database, and have the priority levels set such that the data suck won't affect production users/usage, then again, that bottleneck is by design.
I'm taking that statement as something that's a comment on the system, rather than a slam on Oracle's technology and capabilities.
Yep, my post was not meant to slam Oracle in any way. It only meant to outline that our Oracle instance was being taxed such that the data pull could not be as aggressive as I would have liked.
Right, I would have thought it impressive if any major DB was the source. Getting it to the point where the source DB is taxed as much as possible seems to be the point of all your optimization. Nice that you could get there.
As a current consumer of an Oracle database I can say from the purely anecdotal experience that Oracle is sufficently complicated to set up and run that I'd like to ditch it for just that reason. Our Oracle installation sucks and we are in the process of porting away from it. It's just not worth it to try to tweak it to work better.
Oracle seems to be built with just one goal. Pay the salaries of people who do nothing but consult on Oracle. We run a Production Grade Oracle instance with high quality hardware and it still won't perform as well as an equivalent mysql db does.
(This opinion is purely subjective based on personal experience and should be understood with that in mind. I'm very soured on Oracle right now :-( )
If, for instance, you were pulling this data out of an Oracle instance designed to be a cheap, historical archive, on minimal hardware and cheap disk, then yeah, it can be very limiting in performance.
If you're pulling it out of a production database, and have the priority levels set such that the data suck won't affect production users/usage, then again, that bottleneck is by design.
I'm taking that statement as something that's a comment on the system, rather than a slam on Oracle's technology and capabilities.