Solaris File Systems
Solaris supports a number of different file systems some out of the box, others by ways of additional software. I will share with you a little about some of the file systems.
In a nutshell Solaris' fie systems have by default been UFS based, however they have partnered with people like Veritas and utilised VxFS as an option and I am working for Sun Microsystems since they've acquired LSC bring the QFS file system into the family. QFS and it's archival product SAM is where my journey began outside of the offline storage arena (that's a story for another day).
Back on track,VxFS from Veritas, and Sun's QFS provided a number of advantages over UFS, the biggest advantage was the ability to cater for extremely large files and file systems. With UFS the maximum for files and file systems is a terabyte, whilst VxFS and QFS offered petrabytes.
The following table gives a small sample of regular disk-based file systems:-
| FS Name | Description |
|---|---|
| UFS | The Berkeley fat fast file system (default for Solaris) |
| VxFS | Veritas file system |
| QFS | LCS's file system |
| PCFS | MS-DOS fat file system, usually associated with floppy dsks |
| HSFS | High Sierra file system, usually associated with CD-ROMs |
| UDFS | Universal Disk Fomat, associated with DVDs |
| TMPFS | Temporary file system (usually memory based and used for swapping |
For the 3 file systems I work with, this table provides a small comparison
| Option | UFS | VxFS | QFS |
|---|---|---|---|
| Maximum file size | 1TB | 8,000 TB | 1PB |
| Maximum file system size | 1TB | 8,000 TB | 1PB |
| Allocation based method | Block | Extent | Extent |
| ACL support | Y | Y | N |
| HSM Support | N | Y | Y |
| Logging | Y (metadta) | Y (metadata) | N |
| Direct I/O | Y (Solaris 2.6 onwards) | Y | Y |
| Quotas | Y | Y | N |
| Page Cache | Y | N | N |
Whilst differences exist between UFS and it's high-capacity neighbours, not just whether the file system is block or extent based, but whether files are extended a block at a time or a number of blocks at a time. Another difference recently raises it's head and that is the support for hierarchal storage management (HSM) in which data flows transparently between faster online and slower near-line media when it has not been recently accessed. HSM although not available for UFS works very well with VXFS and QFS.
if you need a file system that is greater than 1TB or maybe that you require additional functionality/features then you may want to read this interesting article which has an insight into trade-offs for using these and other file systems.
PCFS provided the means and support for the implementation for MS-DOS FAT file system for use mainly with floppy disks.
HSFS (aka ISO 9660) commonly associated CD-ROMs. A bonus under Solaris is that HSFS also supported the Rock Ridge extentions to provided unix-style file naming.
With the introduction of Solaris 8 we see the support for UDFs namly support for DVD devices.
TMPFS may look and act like a disk-based file system, but they are actually resident in memory.
The primary goal o /tmp is for memory swappig. IE memory is not written to disk unless physical memory is exhausted, therefore /tmp are only temporary storage areas.. I'll discuss swapping a little later in another post.
NFS (the network file system) common across almost all operating systems provides a file-sharing mechanism across a network and across different platforms.
[part 2 - pseudo file systems]