git.m6w6.name Git - awesomized/libmemcached/blob - bin/memaslap.rst

   1 ==================================================
   2 memaslap - Load testing  and benchmarking a server
   3 ==================================================
   4
   5 SYNOPSIS
   6 --------
   7
   8 memaslap [options]
   9
  10 .. program:: memaslap
  11
  12 .. option:: --help
  13
  14 .. envvar:: MEMCACHED_SERVERS
  15
  16 DESCRIPTION
  17 -----------
  18
  19 :program:`memaslap` is a load generation and benchmark tool for memcached
  20 servers. It generates configurable workload such as threads, concurrency,
  21 connections, run time, overwrite, miss rate, key size, value size, get/set
  22 proportion, expected throughput, and so on. Furthermore, it also tests data
  23 verification, expire-time verification, UDP, binary protocol, facebook test,
  24 replication test, multi-get and reconnection, etc.
  25
  26 Memaslap manages network connections like memcached with
  27 libevent. Each thread of memaslap is bound with a CPU core, all
  28 the threads don't communicate with each other, and there are several socket
  29 connections in each thread. Each connection keeps key size distribution,
  30 value size distribution, and command distribution by itself.
  31
  32 You can specify servers via the :option:`memslap --servers` option or via the
  33 environment variable :envvar:`MEMCACHED_SERVERS`.
  34
  35 FEATURES
  36 --------
  37
  38 Memslap is developed to for the following purposes:
  39
  40 Manages network connections with libevent asynchronously.
  41
  42 Set both TCP and UDP up to use non-blocking IO.
  43
  44 Improves parallelism: higher performance in multi-threads environments.
  45
  46 Improves time efficiency: faster processing speed.
  47
  48 Generates key and value more efficiently; key size distribution and value size distribution are configurable.
  49
  50 Supports get, multi-get, and set commands; command distribution is configurable.
  51
  52 Supports controllable miss rate and overwrite rate.
  53
  54 Supports data and expire-time verification.
  55
  56 Supports dumping statistic information periodically.
  57
  58 Supports thousands of TCP connections.
  59
  60 Supports binary protocol.
  61
  62 Supports facebook test (set with TCP and multi-get with UDP) and replication test.
  63
  64 DETAILS
  65 -------
  66
  67 Effective implementation of network.
  68 ____________________________________
  69
  70 For memaslap, both TCP and UDP use non-blocking network IO. All
  71 the network events are managed by libevent as memcached. The network module
  72 of memaslap is similar to memcached. Libevent can ensure
  73 memaslap can handle network very efficiently.
  74
  75 Effective implementation of multi-threads and concurrency
  76 _________________________________________________________
  77
  78 Memslap has the similar implementation of multi-threads to
  79 memcached. Memslap creates one or more self-governed threads;
  80 each thread is bound with one CPU core if the system tests setting CPU
  81 core affinity.
  82
  83 In addition, each thread has a libevent to manage the events of the network;
  84 each thread has one or more self-governed concurrency; and each
  85 concurrency has one or more socket connections. All the concurrent tasks don't
  86 communicate with each other even though they are in the same thread.
  87
  88 Memslap can create thousands of socket connections, and each
  89 concurrency has tens of socket connections. Each concurrency randomly or
  90 sequentially selects one socket connection from its socket connection pool
  91 to run, so memaslap can ensure each concurrency handles one
  92 socket connection at any given time. Users can specify the number of
  93 concurrency and socket connections of each concurrency according to their
  94 expected workload.
  95
  96 Effective implementation of generating key and value
  97 ____________________________________________________
  98
  99 In order to improve time efficiency and space efficiency,
 100 memaslap creates a random characters table with 10M characters. All the
 101 suffixes of keys and values are generated from this random characters table.
 102
 103 Memslap uses the offset in the character table and the length
 104 of the string to identify a string. It can save much memory.
 105 Each key contains two parts, a prefix and a suffix. The prefix is an
 106 uint64_t, 8 bytes. In order to verify the data set before,
 107 memaslap need to ensure each key is unique, so it uses the prefix to identify
 108 a key. The prefix cannot include illegal characters, such as '\r', '\n',
 109 '\0' and ' '. And memaslap has an algorithm to ensure that.
 110
 111 Memslap doesn't generate all the objects (key-value pairs) at
 112 the beginning. It only generates enough objects to fill the task window
 113 (default 10K objects) of each concurrency. Each object has the following
 114 basic information, key prefix, key suffix offset in the character table, key
 115 length, value offset in the character table, and value length.
 116
 117 In the work process, each concurrency sequentially or randomly selects an
 118 object from the window to do set operation or get operation. At the same
 119 time, each concurrency kicks objects out of its window and adds new object
 120 into it.
 121
 122 Simple but useful task scheduling
 123 _________________________________
 124
 125 Memslap uses libevent to schedule all concurrent tasks of
 126 threads, and each concurrency schedules tasks based on the local task
 127 window. Memslap assumes that if each concurrency keeps the same
 128 key distribution, value distribution and commands distribution, from
 129 outside, memaslap keeps all the distribution as a whole.
 130 Each task window includes a lot of objects, each object stores its basic
 131 information, such as key, value, expire time, and so on. At any time, all
 132 the objects in the window keep the same and fixed key and value
 133 distribution. If an object is overwritten, the value of the object will be
 134 updated. Memslap verifies the data or expire-time according to
 135 the object information stored in the task window.
 136
 137 Libevent selects which concurrency to handle based on a specific network
 138 event. Then the concurrency selects which command (get or set) to operate
 139 based on the command distribution. If it needs to kick out an old object and
 140 add a new object, in order to keep the same key and value distribution, the
 141 new object must have the same key length and value length.
 142
 143 If memcached server has two cache layers (memory and SSD), running
 144 memaslap with different window sizes can get different cache
 145 miss rates. If memaslap adds enough objects into the windows at
 146 the beginning, and the cache of memcached cannot store all the objects
 147 initialized, then memaslap will get some objects from the second
 148 cache layer. It causes the first cache layer to miss. So the user can
 149 specify the window size to get the expected miss rate of the first cache
 150 layer.
 151
 152 Useful implementation of multi-servers , UDP, TCP, multi-get and binary protocol
 153 ________________________________________________________________________________
 154
 155 Because each thread is self-governed, memaslap can assign
 156 different threads to handle different memcached servers. This is just one of
 157 the ways in which memaslap tests multiple servers. The only
 158 limitation is that the number of servers cannot be greater than the number
 159 of threads. The other way to test multiple servers is for replication
 160 test. Each concurrency has one socket connection to each memcached server.
 161 For the implementation, memaslap can set some objects to one
 162 memcached server, and get these objects from the other servers.
 163
 164 By default, Memslap does single get. If the user specifies
 165 multi-get option, memaslap will collect enough get commands and
 166 pack and send the commands together.
 167
 168 Memslap tests both the ASCII protocol and binary protocol,
 169 but it runs on the ASCII protocol by default.
 170 Memslap by default runs on the TCP protocol, but it also
 171 tests UDP. Because UDP is unreliable, dropped packages and out-of-order
 172 packages may occur. Memslap creates a memory buffer to handle
 173 these problems. Memslap tries to read all the response data of
 174 one command from the server and reorders the response data. If some packages
 175 get lost, the waiting timeout mechanism can ensure half-baked packages will
 176 be discarded and the next command will be sent.
 177
 178 USAGE
 179 -----
 180
 181 Below are some usage samples:
 182
 183 memaslap -s 127.0.0.1:11211 -S 5s
 184
 185 memaslap -s 127.0.0.1:11211 -t 2m -v 0.2 -e 0.05 -b
 186
 187 memaslap -s 127.0.0.1:11211 -F config -t 2m -w 40k -S 20s -o 0.2
 188
 189 memaslap -s 127.0.0.1:11211 -F config -t 2m -T 4 -c 128 -d 20 -P 40k
 190
 191 memaslap -s 127.0.0.1:11211 -F config -t 2m -d 50 -a -n 40
 192
 193 memaslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m
 194
 195 memaslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m -p 2
 196
 197 The user must specify one server at least to run memaslap. The
 198 rest of the parameters have default values, as shown below:
 199
 200 Thread number = 1                     Concurrency = 16
 201
 202 Run time = 600 seconds                Configuration file = NULL
 203
 204 Key size = 64                         Value size = 1024
 205
 206 Get/set = 9:1                         Window size = 10k
 207
 208 Execute number = 0                    Single get = true
 209
 210 Multi-get = false                     Number of sockets of each concurrency = 1
 211
 212 Reconnect = false                     Data verification = false
 213
 214 Expire-time verification = false      ASCII protocol = true
 215
 216 Binary protocol = false               Dumping statistic information periodically = false
 217
 218 Overwrite proportion = 0%             UDP = false
 219
 220 TCP = true                            Limit throughput = false
 221
 222 Facebook test = false                 Replication test = false
 223
 224 Key size, value size and command distribution.
 225 ______________________________________________
 226
 227 All the distributions are read from the configuration file specified by user
 228 with "—cfg_cmd" option. If the user does not specify a configuration file,
 229 memaslap will run with the default distribution (key size = 64,
 230 value size = 1024, get/set = 9:1). For information on how to edit the
 231 configuration file, refer to the "Configuration File" section.
 232
 233 The minimum key size is 16 bytes; the maximum key size is 250 bytes. The
 234 precision of proportion is 0.001. The proportion of distribution will be
 235 rounded to 3 decimal places.
 236
 237 The minimum value size is 1 bytes; the maximum value size is 1M bytes. The
 238 precision of proportion is 0.001. The proportion of distribution will be
 239 rounded to 3 decimal places.
 240 Currently, memaslap only tests set and get commands. And it
 241 testss 100% set and 100% get. For 100% get, it will preset some objects to
 242 the server.
 243
 244 Multi-thread and concurrency
 245 ____________________________
 246
 247 The high performance of memaslap benefits from the special
 248 schedule of thread and concurrency. It's important to specify the proper
 249 number of them. The default number of threads is 1; the default number of
 250 concurrency is 16. The user can use "—threads" and "--concurrency" to
 251 specify these variables.
 252
 253 If the system tests setting CPU affinity and the number of threads
 254 specified by the user is greater than 1, memaslap will try to
 255 bind each thread to a different CPU core. So if you want to get the best
 256 performance memaslap, it is better to specify the number of
 257 thread equal to the number of CPU cores. The number of threads specified by
 258 the user can also be less or greater than the number of CPU cores. Because
 259 of the limitation of implementation, the number of concurrencies could be
 260 the multiple of the number of threads.
 261
 262 1. For 8 CPU cores system
 263
 264 For example:
 265
 266 --threads=2 --concurrency=128
 267
 268 --threads=8 --concurrency=128
 269
 270 --threads=8 --concurrency=256
 271
 272 --threads=12 --concurrency=144
 273
 274 2. For 16 CPU cores system
 275
 276 For example:
 277
 278 --threads=8 --concurrency=128
 279
 280 --threads=16 --concurrency=256
 281
 282 --threads=16 --concurrency=512
 283
 284 --threads=24 --concurrency=288
 285
 286 The memaslap performs very well, when
 287 used to test the performance of memcached servers.
 288 Most of the time, the bottleneck is the network or
 289 the server. If for some reason the user wants to
 290 limit the performance of memaslap, there
 291 are two ways to do this:
 292
 293 Decrease the number of threads and concurrencies.
 294 Use the option "--tps" that memaslap
 295 provides to limit the throughput. This option allows
 296 the user to get the expected throughput. For
 297 example, assume that the maximum throughput is 50
 298 kops/s for a specific configuration, you can specify
 299 the throughput equal to or less than the maximum
 300 throughput using "--tps" option.
 301
 302 Window size
 303 ___________
 304
 305 Most of the time, the user does not need to specify the window size. The
 306 default window size is 10k. For Schooner Memcached, the user can specify
 307 different window sizes to get different cache miss rates based on the test
 308 case. Memslap testss cache miss rate between 0% and 100%.
 309 If you use this utility to test the performance of Schooner Memcached, you
 310 can specify a proper window size to get the expected cache miss rate. The
 311 formula for calculating window size is as follows:
 312
 313 Assume that the key size is 128 bytes, and the value size is 2048 bytes, and
 314 concurrency=128.
 315
 316 1. Small cache cache_size=1M, 100% cache miss (all data get from SSD).
 317 win_size=10k
 318
 319 2. cache_size=4G
 320
 321 (1). cache miss rate 0%
 322
 323 win_size=8k
 324
 325 (2). cache miss rate 5%
 326
 327 win_size=11k
 328
 329 3. cache_size=16G
 330
 331 (1). cache miss rate 0%
 332
 333 win_size=32k
 334
 335 (2). cache miss
 336
 337 rate 5%
 338
 339 win_size=46k
 340
 341 The formula for calculating window size for cache miss rate 0%:
 342
 343 cache_size / concurrency / (key_size + value_size) \* 0.5
 344
 345 The formula for calculating window size for cache miss rate 5%:
 346
 347 cache_size / concurrency / (key_size + value_size) \* 0.7
 348
 349 Verification
 350 ____________
 351
 352 Memslap testss both data verification and expire-time
 353 verification. The user can use "--verify=" or "-v" to specify the proportion
 354 of data verification. In theory, it testss 100% data verification. The
 355 user can use "--exp_verify=" or "-e" to specify the proportion of
 356 expire-time verification. In theory, it testss 100% expire-time
 357 verification. Specify the "--verbose" options to get more detailed error
 358 information.
 359
 360 For example: --exp_verify=0.01 –verify=0.1 , it means that 1% of the objects
 361 set with expire-time, 10% of the objects gotten will be verified. If the
 362 objects are gotten, memaslap will verify the expire-time and
 363 value.
 364
 365 multi-servers and multi-config
 366 _______________________________
 367
 368 Memslap testss multi-servers based on self-governed thread.
 369 There is a limitation that the number of servers cannot be greater than the
 370 number of threads. Memslap assigns one thread to handle one
 371 server at least. The user can use the "--servers=" or "-s" option to specify
 372 multi-servers.
 373
 374 For example:
 375
 376 --servers=10.1.1.1:11211,10.1.1.2:11212,10.1.1.3:11213 --threads=6 --concurrency=36
 377
 378 The above command means that there are 6 threads, with each thread having 6
 379 concurrencies and that threads 0 and 3 handle server 0 (10.1.1.1); threads 1
 380 and 4 handle server 1 (10.1.1.2); and thread 2 and 5 handle server 2
 381 (10.1.1.3).
 382
 383 All the threads and concurrencies in memaslap are self-governed.
 384
 385 So is memaslap. The user can start up several
 386 memaslap instances. The user can run memaslap on different client
 387 machines to communicate with the same memcached server at the same. It is
 388 recommended that the user start different memaslap on different
 389 machines using the same configuration.
 390
 391 Run with execute number mode or time mode
 392 _________________________________________
 393
 394 The default memaslap runs with time mode. The default run time
 395 is 10 minutes. If it times out, memaslap will exit. Do not
 396 specify both execute number mode and time mode at the same time; just
 397 specify one instead.
 398
 399 For example:
 400
 401 --time=30s (It means the test will run 30 seconds.)
 402
 403 --execute_number=100000 (It means that after running 100000 commands, the test will exit.)
 404
 405 Dump statistic information periodically.
 406 ________________________________________
 407
 408 The user can use "--stat_freq=" or "-S" to specify the frequency.
 409
 410 For example:
 411
 412 --stat_freq=20s
 413
 414 Memslap will dump the statistics of the commands (get and set) at the frequency of every 20
 415 seconds.
 416
 417 For more information on the format of dumping statistic information, refer to "Format of Output" section.
 418
 419 Multi-get
 420 _________
 421
 422 The user can use "--division=" or "-d" to specify multi-get keys count.
 423 Memslap by default does single get with TCP. Memslap also testss data
 424 verification and expire-time verification for multi-get.
 425
 426 Memslap testss multi-get with both TCP and UDP. Because of
 427 the different implementation of the ASCII protocol and binary protocol,
 428 there are some differences between the two. For the ASCII protocol,
 429 memaslap sends one "multi-get" to the server once. For the
 430 binary protocol, memaslap sends several single get commands
 431 together as "multi-get" to the server.
 432
 433 UDP and TCP
 434 ___________
 435
 436 Memslap testss both UDP and TCP. For TCP,
 437 memaslap does not reconnect the memcached server if socket connections are
 438 lost. If all the socket connections are lost or memcached server crashes,
 439 memaslap will exit. If the user specifies the "--reconnect"
 440 option when socket connections are lost, it will reconnect them.
 441
 442 User can use "--udp" to enable the UDP feature, but UDP comes with some
 443 limitations:
 444
 445 UDP cannot set data more than 1400 bytes.
 446
 447 UDP is not tested by the binary protocol because the binary protocol of
 448 memcached does not tests that.
 449
 450 UDP doesn't tests reconnection.
 451
 452 Facebook test
 453 _____________
 454
 455 Set data with TCP and multi-get with UDP. Specify the following options:
 456
 457 "--facebook --division=50"
 458
 459 If you want to create thousands of TCP connections, specify the
 460
 461 "--conn_sock=" option.
 462
 463 For example: --facebook --division=50 --conn_sock=200
 464
 465 The above command means that memaslap will do facebook test,
 466 each concurrency has 200 socket TCP connections and one UDP socket.
 467
 468 Memslap sets objects with the TCP socket, and multi-gets 50
 469 objects once with the UDP socket.
 470
 471 If you specify "--division=50", the key size must be less that 25 bytes
 472 because the UDP packet size is 1400 bytes.
 473
 474 Replication test
 475 ________________
 476
 477 For replication test, the user must specify at least two memcached servers.
 478 The user can use "—rep_write=" option to enable feature.
 479
 480 For example:
 481
 482 --servers=10.1.1.1:11211,10.1.1.2:11212 –rep_write=2
 483
 484 The above command means that there are 2 replication memcached servers,
 485 memaslap will set objects to both server 0 and server 1, get
 486 objects which are set to server 0 before from server 1, and also get objects
 487 which are set to server 1 before from server 0. If server 0 crashes,
 488 memaslap will only get objects from server 1. If server 0 comes
 489 back to life again, memaslap will reconnect server 0. If both
 490 server 0 and server 1 crash, memaslap will exit.
 491
 492 Supports thousands of TCP connections
 493 _____________________________________
 494
 495 Start memaslap with "--conn_sock=" or "-n" to enable this
 496 feature. Make sure that your system can tests opening thousands of files
 497 and creating thousands of sockets. However, this feature does not tests
 498 reconnection if sockets disconnect.
 499
 500 For example:
 501
 502 --threads=8 --concurrency=128 --conn_sock=128
 503
 504 The above command means that memaslap starts up 8 threads, each
 505 thread has 16 concurrencies, each concurrency has 128 TCP socket
 506 connections, and the total number of TCP socket connections is 128 \* 128 =
 507 16384.
 508
 509 Supports binary protocol
 510 ________________________
 511
 512 Start memaslap with "--binary" or "-B" options to enable this
 513 feature. It testss all the above features except UDP, because the latest
 514 memcached 1.3.3 does not implement binary UDP protocol.
 515
 516 For example:
 517
 518 --binary
 519
 520 Since memcached 1.3.3 doesn't implement binary UDP protocol,
 521 memaslap does not tests UDP. In addition, memcached 1.3.3 does not tests
 522 multi-get. If you specify "--division=50" option, it just sends 50 get
 523 commands together as "multi-get" to the server.
 524
 525 Configuration file
 526 ------------------
 527
 528 This section describes the format of the configuration file.  By default
 529 when no configuration file is specified memaslap reads the default
 530 one located at ~/.memaslap.cnf.
 531
 532 Below is a sample configuration file:
 533
 534 .. code-block:: perl
 535
 536   ---------------------------------------------------------------------------
 537   #comments should start with '#'
 538   #key
 539   #start_len end_len proportion
 540   #
 541   #key length range from start_len to end_len
 542   #start_len must be equal to or greater than 16
 543   #end_len must be equal to or less than 250
 544   #start_len must be equal to or greater than end_len
 545   #memaslap will generate keys according to the key range
 546   #proportion: indicates keys generated from one range accounts for the total
 547   generated keys
 548   #
 549   #example1: key range 16~100 accounts for 80%
 550   #          key range 101~200 accounts for 10%
 551   #          key range 201~250 accounts for 10%
 552   #          total should be 1 (0.8+0.1+0.1 = 1)
 553   #
 554   #          16 100 0.8
 555   #          101 200 0.1
 556   #          201 249 0.1
 557   #
 558   #example2: all keys length are 128 bytes
 559   #
 560   #          128 128 1
 561   key
 562   128 128 1
 563   #value
 564   #start_len end_len proportion
 565   #
 566   #value length range from start_len to end_len
 567   #start_len must be equal to or greater than 1
 568   #end_len must be equal to or less than 1M
 569   #start_len must be equal to or greater than end_len
 570   #memaslap will generate values according to the value range
 571   #proportion: indicates values generated from one range accounts for the
 572   total generated values
 573   #
 574   #example1: value range 1~1000 accounts for 80%
 575   #          value range 1001~10000 accounts for 10%
 576   #          value range 10001~100000 accounts for 10%
 577   #          total should be 1 (0.8+0.1+0.1 = 1)
 578   #
 579   #          1 1000 0.8
 580   #          1001 10000 0.1
 581   #          10001 100000 0.1
 582   #
 583   #example2: all value length are 128 bytes
 584   #
 585   #          128 128 1
 586   value
 587   2048 2048 1
 588   #cmd
 589   #cmd_type cmd_proportion
 590   #
 591   #currently memaslap only testss get and set command.
 592   #
 593   #cmd_type
 594   #set     0
 595   #get     1
 596   #
 597   #example: set command accounts for 50%
 598   #         get command accounts for 50%
 599   #         total should be 1 (0.5+0.5 = 1)
 600   #
 601   #         cmd
 602   #         0    0.5
 603   #         1    0.5
 604   cmd
 605   0    0.1
 606   1.0 0.9
 607
 608 Format of output
 609 ----------------
 610
 611 At the beginning, memaslap displays some configuration information as follows:
 612
 613 servers : 127.0.0.1:11211
 614
 615 threads count: 1
 616
 617 concurrency: 16
 618
 619 run time: 20s
 620
 621 windows size: 10k
 622
 623 set proportion: set_prop=0.10
 624
 625 get proportion: get_prop=0.90
 626
 627 Where
 628 _____
 629
 630 servers : "servers"
 631
 632  The servers used by memaslap.
 633
 634 threads count
 635
 636  The number of threads memaslap runs with.
 637
 638 concurrency
 639
 640  The number of concurrencies memaslap runs with.
 641
 642 run time
 643
 644  How long to run memaslap.
 645
 646 windows size
 647
 648  The task window size of each concurrency.
 649
 650 set proportion
 651
 652  The proportion of set command.
 653
 654 get proportion
 655
 656  The proportion of get command.
 657
 658 The output of dynamic statistics is something like this:
 659
 660 .. code-block:: perl
 661
 662   ---------------------------------------------------------------------------------------------------------------------------------
 663   Get Statistics
 664   Type  Time(s)  Ops   TPS(ops/s)  Net(M/s)  Get_miss  Min(us)  Max(us)
 665   Avg(us)  Std_dev    Geo_dist
 666   Period   5   345826  69165     65.3      0         27      2198     203
 667   95.43      177.29
 668   Global  20  1257935  62896     71.8      0         26      3791     224
 669   117.79     192.60
 670   Set Statistics
 671
 672   Type  Time(s)  Ops   TPS(ops/s)  Net(M/s)  Get_miss  Min(us)  Max(us)
 673   Avg(us)  Std_dev    Geo_dist
 674   Period   5    38425   7685      7.3       0         42      628     240
 675   88.05      220.21
 676   Global   20   139780  6989      8.0       0         37      3790    253
 677   117.93     224.83
 678   Total Statistics
 679
 680   Type  Time(s)  Ops   TPS(ops/s)  Net(M/s)  Get_miss  Min(us)  Max(us)
 681   Avg(us)  Std_dev    Geo_dist
 682   Period   5   384252   76850     72.5      0        27      2198     207
 683   94.72      181.18
 684   Global  20  1397720   69886     79.7      0        26      3791     227
 685   117.93     195.60
 686   ---------------------------------------------------------------------------------------------------------------------------------
 687
 688 Where
 689 _____
 690
 691 Get Statistics
 692
 693  Statistics information of get command
 694
 695 Set Statistics
 696
 697  Statistics information of set command
 698
 699 Total Statistics
 700
 701  Statistics information of both get and set command
 702
 703 Period
 704
 705  Result within a period
 706
 707 Global
 708
 709  Accumulated results
 710
 711 Ops
 712
 713  Total operations
 714
 715 TPS
 716
 717  Throughput, operations/second
 718
 719 Net
 720
 721  The rate of network
 722
 723 Get_miss
 724
 725  How many objects can't be gotten
 726
 727 Min
 728
 729  The minimum response time
 730
 731 Max
 732
 733  The maximum response time
 734
 735 Avg:
 736
 737  The average response time
 738
 739 Std_dev
 740
 741  Standard deviation of response time
 742
 743 Geo_dist
 744
 745  Geometric distribution based on natural exponential function
 746
 747 At the end, memaslap will output something like this:
 748
 749 .. code-block:: perl
 750
 751    ---------------------------------------------------------------------------------------------------------------------------------
 752    Get Statistics (1257956 events)
 753      Min:        26
 754      Max:      3791
 755      Avg:       224
 756      Geo:    192.60
 757      Std:    116.23
 758                      Log2 Dist:
 759                        4:        0       10    84490   215345
 760                        8:   484890   459823    12543      824
 761                       12:       31
 762
 763     Set Statistics (139782 events)
 764        Min:        37
 765        Max:      3790
 766        Avg:       253
 767        Geo:    224.84
 768        Std:    116.83
 769        Log2 Dist:
 770          4:        0        0     4200 16988
 771          8:    50784    65574 2064      167
 772          12:        5
 773
 774      Total Statistics (1397738 events)
 775          Min:        26
 776          Max:      3791
 777          Avg:       227
 778          Geo:    195.60
 779          Std:    116.60
 780          Log2 Dist:
 781            4:        0       10    88690   232333
 782            8:   535674   525397    14607      991
 783            12:       36
 784
 785    cmd_get: 1257969
 786    cmd_set: 139785
 787    get_misses: 0
 788    verify_misses: 0
 789    verify_failed: 0
 790    expired_get: 0
 791    unexpired_unget: 0
 792    written_bytes: 242516030
 793    read_bytes: 1003702556
 794    object_bytes: 152086080
 795    packet_disorder: 0
 796    packet_drop: 0
 797    udp_timeout: 0
 798
 799    Run time: 20.0s Ops: 1397754 TPS: 69817 Net_rate: 59.4M/s
 800    ---------------------------------------------------------------------------------------------------------------------------------
 801
 802 Where
 803 _____
 804
 805 Get Statistics
 806
 807  Get statistics of response time
 808
 809 Set Statistics
 810
 811  Set statistics of response time
 812
 813 Total Statistics
 814
 815  Both get and set statistics of response time
 816
 817 Min
 818
 819  The accumulated and minimum response time
 820
 821 Max
 822
 823  The accumulated and maximum response time
 824
 825 Avg
 826
 827  The accumulated and average response time
 828
 829 Std
 830
 831  Standard deviation of response time
 832
 833 Log2 Dist
 834
 835  Geometric distribution based on logarithm 2
 836
 837 cmd_get
 838
 839  Total get commands done
 840
 841 cmd_set
 842
 843  Total set commands done
 844
 845 get_misses
 846
 847  How many objects can't be gotten from server
 848
 849 verify_misses
 850
 851  How many objects need to verify but can't get them
 852
 853 verify_failed
 854
 855  How many objects with insistent value
 856
 857 expired_get
 858
 859  How many objects are expired but we get them
 860
 861 unexpired_unget
 862
 863  How many objects are unexpired but we can't get them
 864
 865 written_bytes
 866
 867  Total written bytes
 868
 869 read_bytes
 870
 871  Total read bytes
 872
 873 object_bytes
 874
 875  Total object bytes
 876
 877 packet_disorder
 878
 879  How many UDP packages are disorder
 880
 881 packet_drop
 882
 883  How many UDP packages are lost
 884
 885 udp_timeout
 886
 887  How many times UDP time out happen
 888
 889 Run time
 890
 891  Total run time
 892
 893 Ops
 894
 895  Total operations
 896
 897 TPS
 898
 899  Throughput, operations/second
 900
 901 Net_rate
 902
 903  The average rate of network
 904
 905 OPTIONS
 906 -------
 907
 908 -s, --servers=
 909     List one or more servers to connect. Servers count must be less than
 910     threads count. e.g.: --servers=localhost:1234,localhost:11211
 911
 912 -T, --threads=
 913     Number of threads to startup, better equal to CPU numbers. Default 8.
 914
 915 -c, --concurrency=
 916     Number of concurrency to simulate with load. Default 128.
 917
 918 -n, --conn_sock=
 919     Number of TCP socks per concurrency. Default 1.
 920
 921 -x, --execute_number=
 922     Number of operations(get and set) to execute for the
 923     given test. Default 1000000.
 924
 925 -t, --time=
 926     How long the test to run, suffix: s-seconds, m-minutes, h-hours,
 927     d-days e.g.: --time=2h.
 928
 929 -F, --cfg_cmd=
 930     Load the configure file to get command,key and value distribution list.
 931
 932 -w, --win_size=
 933     Task window size of each concurrency, suffix: K, M e.g.: --win_size=10k.
 934     Default 10k.
 935
 936 -X, --fixed_size=
 937     Fixed length of value.
 938
 939 -v, --verify=
 940     The proportion of date verification, e.g.: --verify=0.01
 941
 942 -d, --division=
 943     Number of keys to multi-get once. Default 1, means single get.
 944
 945 -S, --stat_freq=
 946     Frequency of dumping statistic information. suffix: s-seconds,
 947     m-minutes, e.g.: --resp_freq=10s.
 948
 949 -e, --exp_verify=
 950     The proportion of objects with expire time, e.g.: --exp_verify=0.01.
 951     Default no object with expire time
 952
 953 -o, --overwrite=
 954     The proportion of objects need overwrite, e.g.: --overwrite=0.01.
 955     Default never overwrite object.
 956
 957 -R, --reconnect
 958     Reconnect tests, when connection is closed it will be reconnected.
 959
 960 -U, --udp
 961     UDP tests, default memaslap uses TCP, TCP port and UDP port of
 962     server must be same.
 963
 964 -a, --facebook
 965     Whether it enables facebook test feature, set with TCP and multi-get with UDP.
 966
 967 -B, --binary
 968     Whether it enables binary protocol. Default with ASCII protocol.
 969
 970 -P, --tps=
 971     Expected throughput, suffix: K, e.g.: --tps=10k.
 972
 973 -p, --rep_write=
 974     The first nth servers can write data, e.g.: --rep_write=2.
 975
 976 -b, --verbose
 977     Whether it outputs detailed information when verification fails.
 978
 979 -h, --help
 980     Display this message and then exit.
 981
 982 -V, --version
 983     Display the version of the application and then exit.
 984
 985 EXAMPLES
 986 --------
 987
 988 memaslap -s 127.0.0.1:11211 -S 5s
 989
 990 memaslap -s 127.0.0.1:11211 -t 2m -v 0.2 -e 0.05 -b
 991
 992 memaslap -s 127.0.0.1:11211 -F config -t 2m -w 40k -S 20s -o 0.2
 993
 994 memaslap -s 127.0.0.1:11211 -F config -t 2m -T 4 -c 128 -d 20 -P 40k
 995
 996 memaslap -s 127.0.0.1:11211 -F config -t 2m -d 50 -a -n 40
 997
 998 memaslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m
 999
1000 memaslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m -p 2
1001
1002 SEE ALSO
1003 --------
1004
1005 .. only:: man
1006
1007   :manpage:`memcached(1)` :manpage:`libmemcached(3)`