Hi all,
This is somewhat academic, but I've a curiosity about this. When formatting the disk or disks for a sql-server platform, does it make a difference what cluster size one uses? For example, if the machine will be used, say as a datamart where it's high-volume reads that are important, it seems that you'd want the largest clusters.
However, I recall reading that one of the major dbms's--I thought it was sql-server--is talking to the disk at an even lower level than Windows api--is this true? Ie, does sql server have it's own interface to the disk so when one creates a db, the .mdf file is not so much a file in the OS sense but more like it's own partition? It sounds odd but it also sounds like it might have merit.
Thanks,
--Jim
This is somewhat academic, but I've a curiosity about this. When formatting the disk or disks for a sql-server platform, does it make a difference what cluster size one uses? For example, if the machine will be used, say as a datamart where it's high-volume reads that are important, it seems that you'd want the largest clusters.
However, I recall reading that one of the major dbms's--I thought it was sql-server--is talking to the disk at an even lower level than Windows api--is this true? Ie, does sql server have it's own interface to the disk so when one creates a db, the .mdf file is not so much a file in the OS sense but more like it's own partition? It sounds odd but it also sounds like it might have merit.
Thanks,
--Jim