How to use this backup command

Question

I am learning about solaris backup, I found three useful backup command, I am confusing about them, can not find
explanation on the man page
ufsdump 0f - / | ( cd /cop/0 ; ufsrestore -rf - )
ufsdump 0f - /export/home | ( cd /cop/7 ; ufsrestore -rf - )
installboot /usr/platform/`uname -i`/lib/fs/ufs/bootblk /dev/rdsk/c0t3d0s0
I try first two commands, they are works, my questions are,
1 The hyphen dumps the data to standard output, what is the standard output?
2 why use the ufsrestore -rf -
3 when I finish the first command, there is nothing record in the /etc/dumpdates, why?
4 it seems they are useful to copy from disk to disk, but , when I restore data,
will I use the command ufsrestore as usual?
5 currently, in the real world, what kind of backup command do you use in your
company? Because many books give me command of backuping data to tape,
are you still using tape nowadays? or do you backup data to hard disk instead of
tape?
appreciated for any help, thanks
Edited by: user8117130 on Jul 2, 2011 9:14 PM

Answer 1

Where to begin..
Well, question 1 and 2:
The commands you found are commands which were used to copy the entire root disk, as well as the partition /export/home to a new disk.
Apparently the person who first used them had a new harddrive, which he (or she) partitioned and mounted under /cop/ . He (or she) then copied all the data from the / and /export/home partitions to the new disk.
Hence the use of ufsrestore.
There are two outputs, standard output (also referred to as - or STDOUT) and standard error.
Any normal output from any command (such as 'ls') is sent to STDOUT . You can redirect data from STDOUT to a file by using >, i.e:
ls > /tmp/output.ls
STDERR is where error messages go.
Apart from STDOUT and STDERR you also have something called STDIN (or Standard Input), which is used to feed data to a command.
Just as you can use > to send STDOUT, you can use < to send STDIN, i.e:
cat < /tmp/output.ls
In the example you posted with ufsdump/ufsrestore the use read the content of a partition with ufsdump, then send it through a pipe to ufsrestore, which writes it down at a different place. The result is that the user gets a copy of the / disk under /cop/0 .
As such thats not a backup command, its a way of copying a disk.
Question 3:
Because there is no 'u' option, to update /etc/dumpdates you need to use the u argument to ufsdump, in your example:
ufsdump 0fu - /
Question 4:
Well. There are two ways to do a ufsdump to a disk, one way is to do it the way you did it, by piping it to ufsrestore. In that case you can't restore your files using ufsrestore, since they are already copied to the disk.
The second way to use ufsdump to a disk is to write the entire dump to a file, i.e.:
ufsdump 0fu /path/to/ufsdump_of_root.dmp /
Then you can restore a file using ufsrestore, to use an interactive restorer:
ufsrestore fi /path/to/ufsdump_of_root.dmp
Question 5:
Most companies i have worked at uses a mix, some backups are written to disks and some to tape. In particular the short term backups (in the time frame one week to one month) are disk based and everything else (one month to ten years) are tape-based.
But it really depends on what kind of data it is and what legal requirements you have. If you have to store data for a few years disk backups are not really an option. The benefits of tape backups are that they are quite fast (in particular LTO5), they don't really consume any electricity and you can easily enlarge your archive by buying more tapes.
I once calculated how many harddrives i would need to store 10 years of backups online. I don't remember how many harddrive i needed (but it was 100TB +), but it would require about 10-15kW of electricity...
Harddisk backups was popular a while ago, but i think tape is coming back, at least thats my impression.
.7/M.