RJT> One plus [of a swapfile on a separate dedicated volume] would be that
RJT> you don't have to worry about that swapfile ever getting fragmented. On
RJT> the other hand you're adding a nontrivial seek to that "drive" each time
RJT> the swapfile [volume] needs to be accessed, and fragmentation isn't much
RJT> of a problem for HPFS drives anyhow.
You can add to that the fact that one always specifies the initial allocation
size for the swapfile, and that as a result the HPFS filesystem driver always
attempts to allocate as much contiguous space as it can for the swapfile when
it is first created. (The ability of programs to specify in advance when a
file is being created how large it is likely to be, so that the filesystem
drivers can do fancy tricks with file placement to avoid the necessity for
fragmentation later on, is one of the features of OS/2.) So the possibility
of non-optimal fragmentation is *only* an issue on HPFS if the swapfile
actually has cause to grow whilst the system is running.
¯ JdeBP ®
--- FleetStreet 1.22 NR
633/260
2501/209
* Origin: JdeBP's point, using Squish (2:257/609.3)
|