Changes between Initial Version and Version 1 of iozone


Ignore:
Timestamp:
Mar 17, 2008, 10:34:03 AM (17 years ago)
Author:
wade
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • iozone

    v1 v1  
     1[[PageOutline]]
     2= GPFS Performance Report (Uses bonnie++ command) =
     3
     4rock, wade, 阿碩
     5[[BR]]
     6[[BR]]
     7== Machine Informance ==
     8||Node ||8 nodes (1 server , 7 client provide disks)||
     9||CPU ||Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz (each node)||
     10||Memory ||2GB DDR2 667 (each node)||
     11||Disk  ||320G+160G (each node), All nodes:  (320G+ 160G) * 7 = 3.36T||
     12||NIC ||Intel Corporation 82566DM Gigabit Network Connection||
     13||Switch ||D-link 24 port GE switch||
     14
     15[[BR]]
     16[[BR]]
     17= Command Introduction =
     18
     19== Offical Web ==
     20http://www.iozone.org/ [[BR]]
     21Documentation：[http://www.iozone.org/docs/IOzone_msword_98.pdf PDF]、[http://www.iozone.org/docs/Iozone_ps.gz Postscript gzip]、[http://www.iozone.org/docs/IOzone_msword_98.doc MS Word Doc]
     22
     23[[BR]]
     24== Description ==
     25IOzone是一套檔案系統的 benchmark。
     26
     27[[BR]]
     28== Installation ==
     29{{{
     30sudo apt-get install iozone3
     31}}}
     32
     33[[BR]]
     34== Synopsis ==
     35{{{
     36iozone [-s filesize_Kb] [-r record_size_Kb] [-f [path]filename] [-h]
     37             [-i test] [-E] [-p] [-a] [-A] [-z] [-Z] [-m] [-M] [-t children]
     38             [-l min_number_procs] [-u max_number_procs] [-v] [-R] [-x] [-o]
     39             [-d microseconds] [-F path1 path2...] [-V pattern] [-j stride]
     40             [-T] [-C] [-B] [-D] [-G] [-I] [-H depth] [-k depth] [-U mount_point]
     41             [-S cache_size] [-O] [-L cacheline_size] [-K] [-g maxfilesize_Kb]
     42             [-n minfilesize_Kb] [-N] [-Q] [-P start_cpu] [-e] [-c] [-b Excel.xls]
     43             [-J milliseconds] [-X write_telemetry_filename] [-w] [-W]
     44             [-Y read_telemetry_filename] [-y minrecsize_Kb] [-q maxrecsize_Kb]
     45             [-+u] [-+m cluster_filename] [-+d] [-+x multiplier] [-+p # ]
     46             [-+r] [-+t] [-+X] [-+Z]
     47
     48           -a  Auto mode
     49           -A  Auto2 mode
     50           -b Filename  Create Excel worksheet file
     51           -B  Use mmap() files
     52           -c  Include close in the timing calculations
     53           -C  Show bytes transferred by each child in throughput testing
     54           -d #  Microsecond delay out of barrier
     55           -D  Use msync(MS_ASYNC) on mmap files
     56           -e  Include flush (fsync,fflush) in the timing calculations
     57           -E  Run extension tests
     58           -f filename  to use
     59           -F filenames  for each process/thread in throughput test
     60           -g #  Set maximum file size (in Kbytes) for auto mode (or #m or #g)
     61           -G  Use msync(MS_SYNC) on mmap files
     62           -h  help
     63           -H #  Use POSIX async I/O with # async operations
     64           -i #  Test to run (0=write/rewrite, 1=read/re-read, 2=random-read/write
     65                 3=Read-backwards, 4=Re-write-record, 5=stride-read, 6=fwrite/re-fwrite
     66                 7=fread/Re-fread, 8=random_mix, 9=pwrite/Re-pwrite, 10=pread/Re-pread
     67                 11=pwritev/Re-pwritev, 12=preadv/Re-preadv)
     68           -I  Use VxFS VX_DIRECT, O_DIRECT,or O_DIRECTIO for all file operations
     69           -j #  Set stride of file accesses to (# * record size)
     70           -J #  milliseconds of compute cycle before each I/O operation
     71           -k #  Use POSIX async I/O (no bcopy) with # async operations
     72           -K  Create jitter in the access pattern for readers
     73           -l #  Lower limit on number of processes to run
     74           -L #  Set processor cache line size to value (in bytes)
     75           -m  Use multiple buffers
     76           -M  Report uname -a output
     77           -n #  Set minimum file size (in Kbytes) for auto mode (or #m or #g)
     78           -N  Report results in microseconds per operation
     79           -o  Writes are synch (O_SYNC)
     80           -O  Give results in ops/sec.
     81           -p  Purge on
     82           -P #  Bind processes/threads to processors, starting with this cpu
     83           -q #  Set maximum record size (in Kbytes) for auto mode (or #m or #g)
     84           -Q  Create offset/latency files
     85           -r #  record size in Kb
     86              or -r #k .. size in Kb
     87              or -r #m .. size in Mb
     88              or -r #g .. size in Gb
     89           -R  Generate Excel report
     90           -s #  file size in Kb
     91              or -s #k .. size in Kb
     92              or -s #m .. size in Mb
     93              or -s #g .. size in Gb
     94           -S #  Set processor cache size to value (in Kbytes)
     95           -t #  Number of threads or processes to use in throughput test
     96           -T  Use POSIX pthreads for throughput tests
     97           -u #  Upper limit on number of processes to run
     98           -U  Mount point to remount between tests
     99           -v  version information
     100           -V #  Verify data pattern write/read
     101           -w  Do not unlink temporary file
     102           -W  Lock file when reading or writing
     103           -x  Turn off stone-walling
     104           -X filename  Write telemetry file. Contains lines with (offset reclen compute_time) in ascii
     105           -y #  Set minimum record size (in Kbytes) for auto mode (or #m or #g)
     106           -Y filename  Read  telemetry file. Contains lines with (offset reclen compute_time) in ascii
     107           -z  Used in conjunction with -a to test all possible record sizes
     108           -Z  Enable mixing of mmap I/O and file I/O
     109           -+K Sony special. Manual control of test 8.
     110           -+m  Cluster_filename   Enable Cluster testing
     111           -+d  File I/O diagnostic mode. (To troubleshoot a broken file I/O subsystem)
     112           -+u  Enable CPU utilization output (Experimental)
     113           -+x # Multiplier to use for incrementing file and record sizes
     114           -+p # Percentage of mix to be reads
     115           -+r Enable O_RSYNC|O_SYNC for all testing.
     116           -+t Enable network performance test. Requires -+m
     117           -+n No retests selected.
     118           -+k Use constant aggregate data set size.
     119           -+q Delay in seconds between tests.
     120           -+l Enable record locking mode.
     121           -+L Enable record locking mode, with shared file.
     122           -+B Sequential mixed workload.
     123           -+D Enable O_DSYNC mode.
     124           -+A #  Enable madvise. 0 = normal, 1=random, 2=sequential
     125                                  3=dontneed, 4=willneed
     126           -+V Enable shared file. No locking.
     127           -+X Enable short circuit mode for filesystem testing ONLY
     128               ALL Results are NOT valid in this mode.
     129           -+Z Enable old data set compatibility mode. WARNING.. Published
     130               hacks may invalidate these results and generate bogus, high
     131               values for results.
     132}}}
     133
     134[[BR]]
     135[[BR]]
     136= GPFS Performance =
     137== Parameters Comments ==
     138 * records sizes =4KB to 16MB
     139 * sizes = 4KB to 16GB
     140
     141== Compare ==
     142|||| 寫入-per char(K/sec) || 寫入-block(K/sec) || 覆寫(K/sec) || 讀取-per char(K/sec) || 讀取-block(K/sec) || 隨機讀取(K/sec) ||
     143|| 8 Nodes, No Replicate, Adjust Parameters ||  58779 || 105491 || 21720 || 50114 || 74658 || 55.4 ||
     144|| 8 Nodes, Replicate, Adjust Parameters ||  52540 || 52879 || 20592 || 53244 || 79776 || 51.4 ||
     145[[Image(GPFS_compare.png)]]