DiskWork
Information about victims
Machine
Disk available
k4
SCSI device sda: 35548320 512-byte hdwr sectors (18201 MB)
Throughput numbers on linux
k4
root@k4:~# uname -a uname -a Linux k4 2.6.11ac6 #4 SMP Wed Apr 6 20:40:08 EDT 2005 ppc64 GNU/Linux root@k4:/proc# dd if=/dev/zero of=/dev/sda bs=4k count=125000 dd if=/dev/zero of=/dev/sda bs=4k count=125000 125000+0 records in 125000+0 records out 512000000 bytes transferred in 20.127782 seconds (25437478 bytes/sec)
k42 experiments
k9, fullDeb, sending requests to disk using async PPCs (delay of 10 ms if EBUSY)
[kix:/u/dilma/k42vanilla] rlogin k9 -l dilma Starting '/bin/bash' on /dev/pts/1 -bash-2.05b$ /kbin/diskTest /dev/scsi/host0/bus0/target9/lun0/part2 125000 /kbin/diskTest /dev/scsi/host0/bus0/target9/lun0/part2 125000 Block size is 4096 numBlocksToWrite is 125000 Time for operations 60.828842 secs (8417059.800078 bytes/sec)
-bash-2.05b$ dd if=/dev/zero of=/dev/scsi/host0/bus0/target9/lun0/part2 bs=4k count=125000 <o of=/dev/scsi/host0/bus0/target9/lun0/part2 bs=4k count=125000 125000+0 records in 125000+0 records out 512000000 bytes transferred in 3269.137000 seconds (156616 bytes/sec) (Don't be scared about how bad this is. This is mainly due to the fact that the device is being treated as a file, and therefore the memory manager will only start issueing write requests once the file is closed! fsync should improve this.)
k9 sending requests to disk using shared memory transport (delay of 10 ms if transport full)
fullDeb-bash-2.05b$ /kbin/diskTest /dev/scsi/host0/bus0/target9/lun0/part6 125000 /kbin/diskTest /dev/scsi/host0/bus0/target9/lun0/part6 125000 Block size is 4096 numBlocksToWrite is 125000 Time for operations 23.514597 secs (21773709.502955 bytes/sec)
noDeb
-bash-2.05b$ /kbin/diskTest /dev/scsi/host0/bus0/target9/lun0/part6 125000 /kbin/diskTest /dev/scsi/host0/bus0/target9/lun0/part6 125000 Block size is 4096 numBlocksToWrite is 125000 Time for operations 21.682900 secs (23613077.065390 bytes/sec)