Re: Swap on invisible partition
- From: Eugen Mezei <eugen.mezei@xxxxxx>
- Date: Sun, 10 Feb 2008 15:08:33 -0800 (PST)
On 10 Feb., 04:41, "John Fullbright" <fjohn@donotspamnetappdotcom>
wrote:
The logical partion makes sense if your server or workstation was built
using an automated build process that starts with a fat partition which is
later converted to NTFS.
The workstations will be installed by deploying NTFS images of the
partitions.
The result
is a boot partion formatted with a 512byte cluster size instead of the
default 4K. This maximizes transactional overhead and negatively impacts
performance. By moving the page file (which uses 4K IOs by the way) to
another logical partion formatted with the default 4K allocation unit size
you can improve performance.
For NT based Windows systems the deployed images contain partitions
that are all formatted with 4K clusters.
If you allow the OS to grow the pagefile, a
logical partion containing only the pagefile will minimize file
fragmentation as well. Of course it's not as good from a performance
perspective as a dedicated spindle, but it is an improvement and the way to
go if you're spindle constrained.
See Also: http://www.tek-tips.com/viewthread.cfm?qid=1399807&page=1
I know a multidisk system is much better optimisable, but as I
explained in the answer to Meinolf, the conditions are given. You must
imagine that I speak here not about a single workstation but about an
entire enterprise computer park, where the workstations are quite the
same configuration with a single harddisk in them. So I can (nor will,
as it would make no sense) not build a second drive, more less a raid
system in every workstation.
In fact what I am concerned about is not to squeeze more performance
out of the workstations by getting a fast swap. (They are already
overpowered for the office applications, with lot of RAM.) What I am
concerned is only how to block access of the users to the partition
where the swapfile resides.
I would like it best if the users would know nothing about the
existence of this partition (that means it doesn't appear to them like
a drive; pretty much like the swappartition in a UNIX system) but the
system will still know how to acces his swapfile.
So: partition hidden from the user but visible to the OS.
Eugen
.
- References:
- Swap on invisible partition
- From: Eugen Mezei
- Re: Swap on invisible partition
- From: Meinolf Weber
- Re: Swap on invisible partition
- From: John Fullbright
- Swap on invisible partition
- Prev by Date: Re: Swap on invisible partition
- Next by Date: Re: FSEXTEND.EXE
- Previous by thread: Re: Swap on invisible partition
- Next by thread: Re: Swap on invisible partition
- Index(es):
Relevant Pages
|
Loading