Skip to content

Instantly share code, notes, and snippets.

@dentarg
Last active December 21, 2015 20:18
Show Gist options
  • Save dentarg/6360181 to your computer and use it in GitHub Desktop.
Save dentarg/6360181 to your computer and use it in GitHub Desktop.
marv vs hild vs Storm SSD
[sda] 1000215216 512-byte logical blocks: (512 GB/476 GiB)
/dev/sda6       413G  2.4G  390G   1% /
$ date ; iozone -l 32 -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 4G ; date                
Fri Nov 22 18:49:51 CET 2013                                                     
        Iozone: Performance Test of File I/O                                     
                Version $Revision: 3.397 $                                       
                Compiled for 64 bit mode.                                        
                Build: linux-AMD64                                               
                                                                                 
        Contributors:William Norcott, Don Capps, Isom Crawford, Kirby Collins    
                     Al Slater, Scott Rhine, Mike Wisner, Ken Goss               
                     Steve Landherr, Brad Smith, Mark Kelly, Dr. Alain CYR,      
                     Randy Dunlap, Mark Montague, Dan Million, Gavin Brebner,    
                     Jean-Marc Zucconi, Jeff Blomberg, Benny Halevy, Dave Boone, 
                     Erik Habbinga, Kris Strecker, Walter Wong, Joshua Root,     
                     Fabrice Bacchella, Zhenghua Xue, Qin Li, Darren Sawyer.     
                     Ben England.                                                
                                                                                 
        Run began: Fri Nov 22 18:49:51 2013                                      
                                                                                 
        OPS Mode. Output is in operations per second.                            
        Include fsync in write timing                                            
        Include fsync in write timing                                            
        No retest option selected                                                
        Record Size 4 KB                                                         
        File size set to 4194304 KB                                              
        Command line used: iozone -l 32 -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 4G     
        Time Resolution = 0.000001 seconds.                                      
        Processor cache size set to 1024 Kbytes.                                 
        Processor cache line size set to 32 bytes.                               
        File stride size set to 17 * record size.                                
        Min process = 32                                                         
        Max process = 32                                                         
        Throughput test with 32 processes                                        
        Each process writes a 4194304 Kbyte file in 4 Kbyte records              
                                                                                 
        Children see throughput for 32 initial writers  =  129293.56 ops/sec     
        Parent sees throughput for 32 initial writers   =  100070.18 ops/sec     
        Min throughput per process                      =    3847.19 ops/sec     
        Max throughput per process                      =    4256.80 ops/sec     
        Avg throughput per process                      =    4040.42 ops/sec     
        Min xfer                                        =  751177.00 ops         
                                                                                 
        Children see throughput for 32 readers          =  137826.06 ops/sec     
        Parent sees throughput for 32 readers           =  137822.04 ops/sec     
        Min throughput per process                      =    4159.09 ops/sec     
        Max throughput per process                      =    5742.21 ops/sec     
        Avg throughput per process                      =    4307.06 ops/sec     
        Min xfer                                        =  759549.00 ops         
                                                                                 
        Children see throughput for 32 random readers   = 1857437.06 ops/sec     
        Parent sees throughput for 32 random readers    = 1852171.47 ops/sec     
        Min throughput per process                      =   11947.84 ops/sec     
        Max throughput per process                      =  404875.16 ops/sec     
        Avg throughput per process                      =   58044.91 ops/sec     
        Min xfer                                        =   30954.00 ops         
                                                                                 
        Children see throughput for 32 random writers   =   91376.89 ops/sec     
        Parent sees throughput for 32 random writers    =   83144.93 ops/sec     
        Min throughput per process                      =    2633.03 ops/sec     
        Max throughput per process                      =    3093.28 ops/sec     
        Avg throughput per process                      =    2855.53 ops/sec     
        Min xfer                                        =  953692.00 ops         
                                                                                 
                                                                                 
                                                                                 
iozone test complete.                                                            
Fri Nov 22 19:04:10 CET 2013                                                     
[sdb] 937703088 512-byte logical blocks: (480 GB/447 GiB)
/dev/sdb        440G  136G  283G  33% /storage
$ date ; iozone -l 32 -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 4G ; date                   
Mon Nov 25 12:52:59 CET 2013                                                        
        Iozone: Performance Test of File I/O                                        
                Version $Revision: 3.397 $                                          
                Compiled for 64 bit mode.                                           
                Build: linux-AMD64                                                  
                                                                                    
        Contributors:William Norcott, Don Capps, Isom Crawford, Kirby Collins       
                     Al Slater, Scott Rhine, Mike Wisner, Ken Goss                  
                     Steve Landherr, Brad Smith, Mark Kelly, Dr. Alain CYR,         
                     Randy Dunlap, Mark Montague, Dan Million, Gavin Brebner,       
                     Jean-Marc Zucconi, Jeff Blomberg, Benny Halevy, Dave Boone,    
                     Erik Habbinga, Kris Strecker, Walter Wong, Joshua Root,        
                     Fabrice Bacchella, Zhenghua Xue, Qin Li, Darren Sawyer.        
                     Ben England.                                                   
                                                                                    
        Run began: Mon Nov 25 12:52:59 2013                                         
                                                                                    
        OPS Mode. Output is in operations per second.                               
        Include fsync in write timing                                               
        No retest option selected                                                   
        Record Size 4 KB                                                            
        File size set to 4194304 KB                                                 
        Command line used: iozone -l 32 -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 4G        
        Time Resolution = 0.000001 seconds.                                         
        Processor cache size set to 1024 Kbytes.                                    
        Processor cache line size set to 32 bytes.                                  
        File stride size set to 17 * record size.                                   
        Min process = 32                                                            
        Max process = 32                                                            
        Throughput test with 32 processes                                           
        Each process writes a 4194304 Kbyte file in 4 Kbyte records                 
                                                                                    
        Children see throughput for 32 initial writers  =   71215.33 ops/sec        
        Parent sees throughput for 32 initial writers   =   12028.68 ops/sec        
        Min throughput per process                      =    1816.28 ops/sec        
        Max throughput per process                      =    7916.33 ops/sec        
        Avg throughput per process                      =    2225.48 ops/sec        
        Min xfer                                        =  161519.00 ops            
                                                                                    
        Children see throughput for 32 readers          =   67352.93 ops/sec        
        Parent sees throughput for 32 readers           =   67349.79 ops/sec        
        Min throughput per process                      =    2092.99 ops/sec        
        Max throughput per process                      =    2107.77 ops/sec        
        Avg throughput per process                      =    2104.78 ops/sec        
        Min xfer                                        = 1041277.00 ops            
                                                                                    
        Children see throughput for 32 random readers   =   26166.23 ops/sec        
        Parent sees throughput for 32 random readers    =   26165.96 ops/sec        
        Min throughput per process                      =     789.18 ops/sec        
        Max throughput per process                      =     844.42 ops/sec        
        Avg throughput per process                      =     817.69 ops/sec        
        Min xfer                                        =  979985.00 ops            
                                                                                    
        Children see throughput for 32 random writers   =   49583.67 ops/sec        
        Parent sees throughput for 32 random writers    =   43947.01 ops/sec        
        Min throughput per process                      =    1357.89 ops/sec        
        Max throughput per process                      =    1698.31 ops/sec        
        Avg throughput per process                      =    1549.49 ops/sec        
        Min xfer                                        =  960330.00 ops            
                                                                                    
                                                                                    
                                                                                    
iozone test complete.                                                               
Mon Nov 25 13:43:13 CET 2013                                                        

iozone: marv vs hild vs Storm SSD

  • marv: MacBook Air 11-inch, Mid 2013 (256 GB APPLE SSD SM0256F)
  • hild: Supermicro SuperServer 2027TR-H72RF+ (6x Samsung 840 Pro, HW RAID10)
  • huor: Supermicro SuperServer 2027TR-H72RF+ (1x Samsung 840 Pro, 1x Samsung SM843)
  • Storm SSD
$ iozone -l 32 -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 4G

    Iozone: Performance Test of File I/O
            Version $Revision: 3.413 $
            Compiled for 64 bit mode.
            Build: macosx

    Contributors:William Norcott, Don Capps, Isom Crawford, Kirby Collins
                 Al Slater, Scott Rhine, Mike Wisner, Ken Goss
                 Steve Landherr, Brad Smith, Mark Kelly, Dr. Alain CYR,
                 Randy Dunlap, Mark Montague, Dan Million, Gavin Brebner,
                 Jean-Marc Zucconi, Jeff Blomberg, Benny Halevy, Dave Boone,
                 Erik Habbinga, Kris Strecker, Walter Wong, Joshua Root,
                 Fabrice Bacchella, Zhenghua Xue, Qin Li, Darren Sawyer.
                 Ben England.

    Run began: Tue Aug 27 23:16:11 2013

    OPS Mode. Output is in operations per second.
    Include fsync in write timing
    No retest option selected
    Record Size 4 KB
    File size set to 4194304 KB
    Command line used: iozone -l 32 -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 4G
    Time Resolution = 0.000001 seconds.
    Processor cache size set to 1024 Kbytes.
    Processor cache line size set to 32 bytes.
    File stride size set to 17 * record size.
    Min process = 32
    Max process = 32
    Throughput test with 32 processes
    Each process writes a 4194304 Kbyte file in 4 Kbyte records

marv

    Children see throughput for 32 initial writers  =  143590.09 ops/sec
    Parent sees throughput for 32 initial writers   =  139070.90 ops/sec
    Min throughput per process                      =    4382.16 ops/sec
    Max throughput per process                      =    4623.34 ops/sec
    Avg throughput per process                      =    4487.19 ops/sec
    Min xfer                                        =  993717.00 ops

    Children see throughput for 32 readers          =  191133.57 ops/sec
    Parent sees throughput for 32 readers           =  190986.70 ops/sec
    Min throughput per process                      =    5830.28 ops/sec
    Max throughput per process                      =    6209.41 ops/sec
    Avg throughput per process                      =    5972.92 ops/sec
    Min xfer                                        =  984575.00 ops

    Children see throughput for 32 random readers   =   40193.81 ops/sec
    Parent sees throughput for 32 random readers    =   40173.70 ops/sec
    Min throughput per process                      =    1250.96 ops/sec
    Max throughput per process                      =    1261.16 ops/sec
    Avg throughput per process                      =    1256.06 ops/sec
    Min xfer                                        = 1040136.00 ops

    Children see throughput for 32 random writers   =   32946.70 ops/sec
    Parent sees throughput for 32 random writers    =   31585.71 ops/sec
    Min throughput per process                      =    1001.99 ops/sec
    Max throughput per process                      =    1068.77 ops/sec
    Avg throughput per process                      =    1029.58 ops/sec
    Min xfer                                        =  984853.00 ops

hild

    Children see throughput for 32 initial writers  =   12285.57 ops/sec
    Parent sees throughput for 32 initial writers   =    1048.07 ops/sec
    Min throughput per process                      =     338.85 ops/sec
    Max throughput per process                      =     748.68 ops/sec
    Avg throughput per process                      =     383.92 ops/sec
    Min xfer                                        =  133745.00 ops

    Children see throughput for 32 readers          = 1059311.75 ops/sec
    Parent sees throughput for 32 readers           = 1059005.89 ops/sec
    Min throughput per process                      =    8649.47 ops/sec
    Max throughput per process                      =   54769.31 ops/sec
    Avg throughput per process                      =   33103.49 ops/sec
    Min xfer                                        =  165629.00 ops

    Children see throughput for 32 random readers   = 7235766.41 ops/sec
    Parent sees throughput for 32 random readers    = 7206898.49 ops/sec
    Min throughput per process                      =    2640.76 ops/sec
    Max throughput per process                      =  649978.50 ops/sec
    Avg throughput per process                      =  226117.70 ops/sec
    Min xfer                                        =    4259.00 ops

    Children see throughput for 32 random writers   =    4318.77 ops/sec
    Parent sees throughput for 32 random writers    =    3950.61 ops/sec
    Min throughput per process                      =     122.19 ops/sec
    Max throughput per process                      =     151.62 ops/sec
    Avg throughput per process                      =     134.96 ops/sec
    Min xfer                                        =  990003.00 ops

Storm SSD

Amazon EC2 m1.large 8GB Storm 12GB Storm SSD
4K Sequential Write 4,900 IOPS 15,000 IOPS 225,000 IOPS
4K Sequential Read 7,800 IOPS 28,000 IOPS 475,000 IOPS
4K Random Write 265 IOPS 204 IOPS 20,000 IOPS
4K Random Read 210 IOPS 700 IOPS 35,000 IOPS

Testing performed using iozone3_398 with the following command line options on each server instance:

iozone -l 32 -O -i 0 -i 1 -i 2 -e -+n -r 4K -s 4G
Amazon EC2 m1.large 8GB Storm 12GB Storm SSD
Sequential Write 36MB/s 110MB/s 900 MB/s
Sequential Read 33MB/s 150MB/s 1,900 MB/s
Random Write 1.5MB/s 1.6MB/s 77 MB/s
Random Read 1.1MB/s 2.4MB/s 142 MB/s

Testing performed using iozone3_398 with the following command line options on each server instance:

iozone -l 32 -i 0 -i 1 -i 2 -e -+n -r 4K -s 2G
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment