Re: A Customer's Request For Open Source Software

[Kilgallen@SpamCop.net (Larry Kilgallen)]

In article <3F53B88E.7040405@attbi.com>, "Robert I. Eachus" <rieachus@attbi.com> writes:

> As I said, most RDBMSs already have the ability to bypass the OS and 
> access disk drives directly.

They do ?  On what operating systems ?

On the one with which I am most familiar, device registers can only be
accessed from the innermost mode, and the code that does that is in a
device driver.  If the DBMS vendor should happen to provide their own
driver, installing it would make it part of the operating system.

Even ADABAS, the most contrary DBMS I have encountered for system manager
relations, was this way.  It insisted on calculating sectors and cylinders
on its own (back when there were such things), but it ended up calling the
operating system's device driver to read those bits into memory.  I think
there are many hardware architectures that would have a hard time delivering
the interrupts without going through the operating system interrupt dispatch.

> This is going one step further, and 
> "mapping" the disk drive into memory.  With a memory-mapped drive on a 
> virtual memory system, you can address the disk just like the memory was 
> physically present in RAM.

I know compilers that read source that way, relying upon the OS paging
in the data as appropriate.  But it is an individual performance/coding
tradeoff, not applied so universally as in Multics.