What's the typical file layout for a large database, and how is the
database accessed?
i.e.
* mostly mmap, or normal read/write calls?
* one huge file, or many independant files?
* do you know if some UNIXes support multiple outstanding read/write
operations per file handle? Under Win32, I can store the file pointer in
an OVERLAPPED structure and send multiple parallel operations to the
same handle. AFAIK, Linux doesn't support that.
* how many parallel read/write operation are attempted per file?
Just one operation, or multiple operations [using multiple filp's?]?
I ask these questions because I try to find an acceptable synchonization
for sys_read()/ sys_write()/ sys_ftruncate()/ O_APPEND. Currently
[2.3.13], this is completely missing, and I'm not sure that i_sem is the
correct sync object.
Regards,
Manfred
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/