[Info-vax] Oracle Database vs Oracle/Rdb

[Jan-Erik Söderholm]

Den 2019-06-28 kl. 04:46, skrev Dave Froble:
> On 6/27/2019 10:22 PM, Arne Vajhøj wrote:
>> On 6/27/2019 10:07 PM, Dave Froble wrote:
>>> On 6/27/2019 5:02 PM, Hein RMS van den Heuvel wrote:
>>>> From a performance perspective, RDb on OpenVMS is probably, but not
>>>> certainly, superior to Oracle-Oracle. They'll probably need more
>>>> capacity (Memory, CPU) to run Oracle-Oracle with similar performance.
>>>
>>> While I'm not a user, one thing has occurred to me.
>>>
>>> Back in the day, when everything ran on one CPU, it didn't matter
>>> which process was running, the CPU could do only one at a time.
>>>
>>> Now today, with lots of CPUs (cores), multiple things can occur
>>> simultaneously, and if every user is waiting on the Oracle engine to
>>> service their needs, it probably is harder to scale that up, than to
>>
>> The Oracle database engine is multi-threaded, so it can service
>> multiple requests in parallel.
>>
>> Arne
>>
>>
> 
> I figured that.  But my question was about scaling up, and that is still my 
> question.  Which is easier and better to scale up?
> 

It is not *that* different. Someone has to schedule the work over the
available cores/CPUs. In Oracle it is done within the Oracle server
process between the threads, in Rdb it is done by the OS between the
user processes.

My main point is that in the Oracle case all database operations must
go through a IPC link between the user/appication process and the DB
"engine". In the Rdb case, each user/application process accesses the
DB files directly. Do not know for sure how much that matters. And note
that it is Rdb that is unique here, most DBs has an architecture that
looks like the one on Oracle...

And in the traditional scale-up scenario using VMS clusters, Rdb is no
issue since it used the DLM. Then there is the usual question about if
it is "better" to scale-up in the box (more/faster cores/CPUs/memory)
or by adding/clustering additional servers. Different pros and cons...