Date: Sat, 15 Jan 2005 14:52:30 +0200 From: Toomas Aas <toomas.aas@raad.tartu.ee> To: questions@freebsd.org Subject: dump performance Message-ID: <41E9120E.7000502@raad.tartu.ee>
next in thread | raw e-mail | index | archive | help
Hello! I'm following the procedure in the NEW-HUGE-DISK FAQ entry to move my system to bigger disks. In a nutshell, the dump/restore combo recommended there seems painfully slow. For example, I have newfs-ed the partition which is to become /usr with the newfs default parameters and mounted it as /mnt/usr. Then I do: # cd /mnt/usr # dump 0af - /usr | restore xf - The procedure starts, and after running for 5 minutes it informs me that dumping is 8.35 % done and should be complete in 54 minutes. This is a total of 1 hour for a partition that holds approximately 2.2 GB of data, making the speed somewhere about 40 MB/minute. Maybe it's just me but it seems way too slow. I tried adding -b 64 and -C 32 switches to dump, and -b 64 to restore. This speeds up the process ca 50 %, but it still means 40 minutes for this partition. 40 minutes is maybe not so bad, but at this speed my /home partition which is 65 GB full would take approximately 18 hours! Tar, on the other hand, did the entire /usr partition in 15 minutes: # cd /mnt # tar cf - -C / usr | tar xf - The source partition is on ATA RAID1 array, consisting of two ATA133 disks attached to Promise TX2 controller. The disks are recognized as ATA133 in dmesg. The destination partition is on an ATA100 disk, attached to ICH4 controller. This disk is recognized as ATA100 in dmesg. Are these speeds normal for dump, or am I missing something?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41E9120E.7000502>