Fix for using correct memcached binary.
[awesomized/libmemcached] / docs / memslap.pod
1 =head1 NAME
2
3 memslap - Load testing and benchmarking tool for memcached
4
5 =head1 SYNOPSIS
6
7 memslap [options]
8
9 =head1 DESCRIPTION
10
11 B<memslap> is a load generation and benchmark tool for memcached(1)
12 servers. It generates configurable workload such as threads, concurrencies, connections,
13 run time, overwrite, miss rate, key size, value size, get/set proportion,
14 expected throughput, and so on. Furthermore, it also supports data
15 verification, expire-time verification, UDP, binary protocol, facebook test,
16 replication test, multi-get and reconnection, etc.
17
18 Memslap manages network connections like memcached with
19 libevent. Each thread of memslap is bound with a CPU core, all
20 the threads don't communicate with each other, and there are several socket
21 connections in each thread. Each connection keeps key size distribution,
22 value size distribution, and command distribution by itself.
23
24 You can specify servers via the B<--servers> option or via the
25 environment variable C<MEMCACHED_SERVERS>.
26
27
28 =head1 FEATURES
29
30 Memslap is developed to for the following purposes:
31
32 =over
33
34 =item Manages network connections with libevent asynchronously.
35
36 =item Set both TCP and UDP up to use non-blocking IO.
37
38 =item Improves parallelism: higher performance in multi-threads environments.
39
40 =item Improves time efficiency: faster processing speed.
41
42 =item Generates key and value more efficiently; key size distribution and value size distribution are configurable.
43
44 =item Supports get, multi-get, and set commands; command distribution is configurable.
45
46 =item Supports controllable miss rate and overwrite rate.
47
48 =item Supports data and expire-time verification.
49
50 =item Supports dumping statistic information periodically.
51
52 =item Supports thousands of TCP connections.
53
54 =item Supports binary protocol.
55
56 =item Supports facebook test (set with TCP and multi-get with UDP) and replication test.
57
58 =back
59
60 =head1 DETAILS
61
62 =head2 Effective implementation of network.
63
64 For memslap, both TCP and UDP use non-blocking network IO. All
65 the network events are managed by libevent as memcached. The network module
66 of memslap is similar to memcached. Libevent can ensure
67 memslap can handle network very efficiently.
68
69 =head2 Effective implementation of multi-threads and concurrency
70
71 Memslap has the similar implementation of multi-threads to
72 memcached. Memslap creates one or more self-governed threads;
73 each thread is bound with one CPU core if the system supports setting CPU
74 core affinity.
75
76 In addition, each thread has a libevent to manage the events of the network;
77 each thread has one or more self-governed concurrencies; and each
78 concurrency has one or more socket connections. All the concurrencies don’t
79 communicate with each other even though they are in the same thread.
80
81 Memslap can create thousands of socket connections, and each
82 concurrency has tens of socket connections. Each concurrency randomly or
83 sequentially selects one socket connection from its socket connection pool
84 to run, so memslap can ensure each concurrency handles one
85 socket connection at any given time. Users can specify the number of
86 concurrency and socket connections of each concurrency according to their
87 expected workload.
88
89 =head2 Effective implementation of generating key and value
90
91 In order to improve time efficiency and space efficiency,
92 memslap creates a random characters table with 10M characters. All the
93 suffixes of keys and values are generated from this random characters table.
94
95 Memslap uses the offset in the character table and the length
96 of the string to identify a string. It can save much memory.
97 Each key contains two parts, a prefix and a suffix. The prefix is an
98 uint64_t, 8 bytes. In order to verify the data set before,
99 memslap need to ensure each key is unique, so it uses the prefix to identify
100 a key. The prefix cannot include illegal characters, such as ‘\r’, ‘\n’,
101 ‘\0’ and ‘ ‘. And memslap has an algorithm to ensure that.
102
103 Memslap doesn’t generate all the objects (key-value pairs) at
104 the beginning. It only generates enough objects to fill the task window
105 (default 10K objects) of each concurrency. Each object has the following
106 basic information, key prefix, key suffix offset in the character table, key
107 length, value offset in the character table, and value length.
108
109 In the work process, each concurrency sequentially or randomly selects an
110 object from the window to do set operation or get operation. At the same
111 time, each concurrency kicks objects out of its window and adds new object
112 into it.
113
114 =head2 Simple but useful task scheduling
115
116 Memslap uses libevent to schedule all the concurrencies of
117 threads, and each concurrency schedules tasks based on the local task
118 window. Memslap assumes that if each concurrency keeps the same
119 key distribution, value distribution and commands distribution, from
120 outside, memslap keeps all the distribution as a whole.
121 Each task window includes a lot of objects, each object stores its basic
122 information, such as key, value, expire time, and so on. At any time, all
123 the objects in the window keep the same and fixed key and value
124 distribution. If an object is overwritten, the value of the object will be
125 updated. Memslap verifies the data or expire-time according to
126 the object information stored in the task window.
127
128 Libevent selects which concurrency to handle based on a specific network
129 event. Then the concurrency selects which command (get or set) to operate
130 based on the command distribution. If it needs to kick out an old object and
131 add a new object, in order to keep the same key and value distribution, the
132 new object must have the same key length and value length.
133
134 If memcached server has two cache layers (memory and SSD), running
135 memslap with different window sizes can get different cache
136 miss rates. If memslap adds enough objects into the windows at
137 the beginning, and the cache of memcached cannot store all the objects
138 initialized, then memslap will get some objects from the second
139 cache layer. It causes the first cache layer to miss. So the user can
140 specify the window size to get the expected miss rate of the first cache
141 layer.
142
143 =head2 Useful implementation of multi-servers , UDP, TCP, multi-get and binary protocol
144
145 Because each thread is self-governed, memslap can assign
146 different threads to handle different memcached servers. This is just one of
147 the ways in which memslap supports multiple servers. The only
148 limitation is that the number of servers cannot be greater than the number
149 of threads. The other way to support multiple servers is for replication
150 test. Each concurrency has one socket connection to each memcached server.
151 For the implementation, memslap can set some objects to one
152 memcached server, and get these objects from the other servers.
153
154 By default, Memslap does single get. If the user specifies
155 multi-get option, memslap will collect enough get commands and
156 pack and send the commands together.
157
158 Memslap supports both the ASCII protocol and binary protocol,
159 but it runs on the ASCII protocol by default.
160 Memslap by default runs on the TCP protocol, but it also
161 supports UDP. Because UDP is unreliable, dropped packages and out-of-order
162 packages may occur. Memslap creates a memory buffer to handle
163 these problems. Memslap tries to read all the response data of
164 one command from the server and reorders the response data. If some packages
165 get lost, the waiting timeout mechanism can ensure half-baked packages will
166 be discarded and the next command will be sent.
167
168
169 =head1 USAGE
170
171 Below are some usage samples:
172
173 =over 4
174
175 =item memslap -s 127.0.0.1:11211 -S 5s
176
177 =item memslap -s 127.0.0.1:11211 -t 2m -v 0.2 -e 0.05 -b
178
179 =item memslap -s 127.0.0.1:11211 -F config -t 2m -w 40k -S 20s -o 0.2
180
181 =item memslap -s 127.0.0.1:11211 -F config -t 2m -T 4 -c 128 -d 20 -P 40k
182
183 =item memslap -s 127.0.0.1:11211 -F config -t 2m -d 50 -a -n 40
184
185 =item memslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m
186
187 =item memslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m -p 2
188
189 =back
190
191 The user must specify one server at least to run memslap. The
192 rest of the parameters have default values, as shown below:
193
194 Thread number = 1 Concurrency = 16
195
196 Run time = 600 seconds Configuration file = NULL
197
198 Key size = 64 Value size = 1024
199
200 Get/set = 9:1 Window size = 10k
201
202 Execute number = 0 Single get = true
203
204 Multi-get = false Number of sockets of each concurrency = 1
205
206 Reconnect = false Data verification = false
207
208 Expire-time verification = false ASCII protocol = true
209
210 Binary protocol = false Dumping statistic information
211
212 periodically = false
213
214 Overwrite proportion = 0% UDP = false
215
216 TCP = true Limit throughput = false
217
218 Facebook test = false Replication test = false
219
220 =head2 Key size, value size and command distribution.
221
222 All the distributions are read from the configuration file specified by user
223 with “—cfg_cmd” option. If the user does not specify a configuration file,
224 memslap will run with the default distribution (key size = 64,
225 value size = 1024, get/set = 9:1). For information on how to edit the
226 configuration file, refer to the “Configuration File” section.
227
228 The minimum key size is 16 bytes; the maximum key size is 250 bytes. The
229 precision of proportion is 0.001. The proportion of distribution will be
230 rounded to 3 decimal places.
231
232 The minimum value size is 1 bytes; the maximum value size is 1M bytes. The
233 precision of proportion is 0.001. The proportion of distribution will be
234 rounded to 3 decimal places.
235 Currently, memslap only supports set and get commands. And it
236 supports 100% set and 100% get. For 100% get, it will preset some objects to
237 the server.
238
239 =head2 Multi-thread and concurrency
240
241 The high performance of memslap benefits from the special
242 schedule of thread and concurrency. It’s important to specify the proper
243 number of them. The default number of threads is 1; the default number of
244 concurrency is 16. The user can use “—threads” and “--concurrency” to
245 specify these variables.
246
247 If the system supports setting CPU affinity and the number of threads
248 specified by the user is greater than 1, memslap will try to
249 bind each thread to a different CPU core. So if you want to get the best
250 performance memslap, it is better to specify the number of
251 thread equal to the number of CPU cores. The number of threads specified by
252 the user can also be less or greater than the number of CPU cores. Because
253 of the limitation of implementation, the number of concurrencies could be
254 the multiple of the number of threads.
255
256 1. For 8 CPU cores system
257
258 For example:
259
260 --threads=2 --concurrency=128
261
262 --threads=8 --concurrency=128
263
264 --threads=8 --concurrency=256
265
266 --threads=12 --concurrency=144
267
268 2. For 16 CPU cores system
269
270 For example:
271
272 --threads=8 --concurrency=128
273
274 --threads=16 --concurrency=256
275
276 --threads=16 --concurrency=512
277
278 --threads=24 --concurrency=288
279
280 The memslap performs very well, when
281 used to test the performance of memcached servers.
282 Most of the time, the bottleneck is the network or
283 the server. If for some reason the user wants to
284 limit the performance of memslap, there
285 are two ways to do this:
286
287 Decrease the number of threads and concurrencies.
288 Use the option “--tps” that memslap
289 provides to limit the throughput. This option allows
290 the user to get the expected throughput. For
291 example, assume that the maximum throughput is 50
292 kops/s for a specific configuration, you can specify
293 the throughput equal to or less than the maximum
294 throughput using “--tps” option.
295
296 =head2 Window size
297
298 Most of the time, the user does not need to specify the window size. The
299 default window size is 10k. For Schooner Memcached, the user can specify
300 different window sizes to get different cache miss rates based on the test
301 case. Memslap supports cache miss rate between 0% and 100%.
302 If you use this utility to test the performance of Schooner Memcached, you
303 can specify a proper window size to get the expected cache miss rate. The
304 formula for calculating window size is as follows:
305
306 Assume that the key size is 128 bytes, and the value size is 2048 bytes, and
307 concurrency=128.
308
309 1. Small cache cache_size=1M, 100% cache miss (all data get from SSD).
310 win_size=10k
311
312 2. cache_size=4G
313
314 (1). cache miss rate 0%
315
316 win_size=8k
317
318 (2). cache miss rate 5%
319
320 win_size=11k
321
322 3. cache_size=16G
323
324 (1). cache miss rate 0%
325
326 win_size=32k
327
328 (2). cache miss
329
330 rate 5%
331
332 win_size=46k
333
334 The formula for calculating window size for cache miss rate 0%:
335
336 cache_size / concurrency / (key_size + value_size) * 0.5
337
338 The formula for calculating window size for cache miss rate 5%:
339
340 cache_size / concurrency / (key_size + value_size) * 0.7
341
342 =head2 Verification
343
344 Memslap supports both data verification and expire-time
345 verification. The user can use "--verify=" or "-v" to specify the proportion
346 of data verification. In theory, it supports 100% data verification. The
347 user can use "--exp_verify=" or "-e" to specify the proportion of
348 expire-time verification. In theory, it supports 100% expire-time
349 verification. Specify the "--verbose" options to get more detailed error
350 information.
351
352 For example: --exp_verify=0.01 –verify=0.1 , it means that 1% of the objects
353 set with expire-time, 10% of the objects gotten will be verified. If the
354 objects are gotten, memslap will verify the expire-time and
355 value.
356
357 =head2 multi-servers and multi-clients
358
359 Memslap supports multi-servers based on self-governed thread.
360 There is a limitation that the number of servers cannot be greater than the
361 number of threads. Memslap assigns one thread to handle one
362 server at least. The user can use the "--servers=" or "-s" option to specify
363 multi-servers.
364
365 For example:
366
367 --servers=10.1.1.1:11211,10.1.1.2:11212,10.1.1.3:11213 --threads=6 --concurrency=36
368
369 The above command means that there are 6 threads, with each thread having 6
370 concurrencies and that threads 0 and 3 handle server 0 (10.1.1.1); threads 1
371 and 4 handle server 1 (10.1.1.2); and thread 2 and 5 handle server 2
372 (10.1.1.3).
373
374 All the threads and concurrencies in memslap are self-governed.
375
376 So is memslap. The user can start up several
377 memslap instances. The user can run memslap on different client
378 machines to communicate with the same memcached server at the same. It is
379 recommended that the user start different memslap on different
380 machines using the same configuration.
381
382 =head2 Run with execute number mode or time mode
383
384 The default memslap runs with time mode. The default run time
385 is 10 minutes. If it times out, memslap will exit. Do not
386 specify both execute number mode and time mode at the same time; just
387 specify one instead.
388
389 For example:
390
391 --time=30s (It means the test will run 30 seconds.)
392
393 --execute_number=100000 (It means that after running 100000 commands, the test will exit.)
394
395 =head2 Dump statistic information periodically.
396
397 The user can use "--stat_freq=" or "-S" to specify the frequency.
398
399 For example:
400
401 --stat_freq=20s
402
403 Memslap will dump the statistics of the commands (get and set) at the frequency of every 20
404 seconds.
405
406 For more information on the format of dumping statistic information, refer to “Format of Output” section.
407
408 =head2 Multi-get
409
410 The user can use "--division=" or "-d" to specify multi-get keys count.
411 Memslap by default does single get with TCP. Memslap also supports data
412 verification and expire-time verification for multi-get.
413
414 Memslap supports multi-get with both TCP and UDP. Because of
415 the different implementation of the ASCII protocol and binary protocol,
416 there are some differences between the two. For the ASCII protocol,
417 memslap sends one “multi-get” to the server once. For the
418 binary protocol, memslap sends several single get commands
419 together as “multi-get” to the server.
420
421 =head2 UDP and TCP
422
423 Memslap supports both UDP and TCP. For TCP,
424 memslap does not reconnect the memcached server if socket connections are
425 lost. If all the socket connections are lost or memcached server crashes,
426 memslap will exit. If the user specifies the “--reconnect”
427 option when socket connections are lost, it will reconnect them.
428
429 User can use “--udp” to enable the UDP feature, but UDP comes with some
430 limitations:
431
432 UDP cannot set data more than 1400 bytes.
433
434 UDP is not supported by the binary protocol because the binary protocol of
435 memcached does not support that.
436
437 UDP doesn’t support reconnection.
438
439 =head2 Facebook test
440
441 Set data with TCP and multi-get with UDP. Specify the following options:
442
443 "--facebook --division=50"
444
445 If you want to create thousands of TCP connections, specify the
446
447 "--conn_sock=" option.
448
449 For example: --facebook --division=50 --conn_sock=200
450
451 The above command means that memslap will do facebook test,
452 each concurrency has 200 socket TCP connections and one UDP socket.
453
454 Memslap sets objects with the TCP socket, and multi-gets 50
455 objects once with the UDP socket.
456
457 If you specify "--division=50", the key size must be less that 25 bytes
458 because the UDP packet size is 1400 bytes.
459
460 =head2 Replication test
461
462 For replication test, the user must specify at least two memcached servers.
463 The user can use “—rep_write=” option to enable feature.
464
465 For example:
466
467 --servers=10.1.1.1:11211,10.1.1.2:11212 –rep_write=2
468
469 The above command means that there are 2 replication memcached servers,
470 memslap will set objects to both server 0 and server 1, get
471 objects which are set to server 0 before from server 1, and also get objects
472 which are set to server 1 before from server 0. If server 0 crashes,
473 memslap will only get objects from server 1. If server 0 comes
474 back to life again, memslap will reconnect server 0. If both
475 server 0 and server 1 crash, memslap will exit.
476
477 =head2 Supports thousands of TCP connections
478
479 Start memslap with "--conn_sock=" or "-n" to enable this
480 feature. Make sure that your system can support opening thousands of files
481 and creating thousands of sockets. However, this feature does not support
482 reconnection if sockets disconnect.
483
484 For example:
485
486 --threads=8 --concurrency=128 --conn_sock=128
487
488 The above command means that memslap starts up 8 threads, each
489 thread has 16 concurrencies, each concurrency has 128 TCP socket
490 connections, and the total number of TCP socket connections is 128 * 128 =
491 16384.
492
493 =head2 Supports binary protocol
494
495 Start memslap with "--binary" or "-B" options to enable this
496 feature. It supports all the above features except UDP, because the latest
497 memcached 1.3.3 does not implement binary UDP protocol.
498
499 For example:
500
501 --binary
502
503 Since memcached 1.3.3 doesn't implement binary UDP protocol,
504 memslap does not support UDP. In addition, memcached 1.3.3 does not support
505 multi-get. If you specify "--division=50" option, it just sends 50 get
506 commands together as “mulit-get” to the server.
507
508 =head1 Configuration file
509
510 This section describes the format of the configuration file. By default
511 when no configuration file is specified memslap reads the default
512 one located at ~/.memslap.cnf.
513
514 Below is a sample configuration file:
515
516 ***************************************************************************
517 #comments should start with '#'
518 #key
519 #start_len end_len proportion
520 #
521 #key length range from start_len to end_len
522 #start_len must be equal to or greater than 16
523 #end_len must be equal to or less than 250
524 #start_len must be equal to or greater than end_len
525 #memslap will generate keys according to the key range
526 #proportion: indicates keys generated from one range accounts for the total
527 generated keys
528 #
529 #example1: key range 16~100 accounts for 80%
530 # key range 101~200 accounts for 10%
531 # key range 201~250 accounts for 10%
532 # total should be 1 (0.8+0.1+0.1 = 1)
533 #
534 # 16 100 0.8
535 # 101 200 0.1
536 # 201 249 0.1
537 #
538 #example2: all keys length are 128 bytes
539 #
540 # 128 128 1
541 key
542 128 128 1
543 #value
544 #start_len end_len proportion
545 #
546 #value length range from start_len to end_len
547 #start_len must be equal to or greater than 1
548 #end_len must be equal to or less than 1M
549 #start_len must be equal to or greater than end_len
550 #memslap will generate values according to the value range
551 #proportion: indicates values generated from one range accounts for the
552 total generated values
553 #
554 #example1: value range 1~1000 accounts for 80%
555 # value range 1001~10000 accounts for 10%
556 # value range 10001~100000 accounts for 10%
557 # total should be 1 (0.8+0.1+0.1 = 1)
558 #
559 # 1 1000 0.8
560 # 1001 10000 0.1
561 # 10001 100000 0.1
562 #
563 #example2: all value length are 128 bytes
564 #
565 # 128 128 1
566 value
567 2048 2048 1
568 #cmd
569 #cmd_type cmd_proportion
570 #
571 #currently memslap only supports get and set command.
572 #
573 #cmd_type
574 #set 0
575 #get 1
576 #
577 #example: set command accounts for 50%
578 # get command accounts for 50%
579 # total should be 1 (0.5+0.5 = 1)
580 #
581 # cmd
582 # 0 0.5
583 # 1 0.5
584 cmd
585 0 0.1
586 1.0 0.9
587
588
589
590 =head1 Format of output
591
592 At the beginning, memslap displays some configuration information as follows:
593
594 =over 4
595
596 =item servers : 127.0.0.1:11211
597
598 =item threads count: 1
599
600 =item concurrency: 16
601
602 =item run time: 20s
603
604 =item windows size: 10k
605
606 =item set proportion: set_prop=0.10
607
608 =item get proportion: get_prop=0.90
609
610 =back
611
612 =head2 Where
613
614 =over 4
615
616 =item servers : "servers"
617
618 The servers used by memslap.
619
620 =item threads count
621
622 The number of threads memslap runs with.
623
624 =item concurrency
625
626 The number of concurrencies memslap runs with.
627
628 =item run time
629
630 How long to run memslap.
631
632 =item windows size
633
634 The task window size of each concurrency.
635
636 =item set proportion
637
638 The proportion of set command.
639
640 =item get proportion
641
642 The proportion of get command.
643
644 =back
645
646 The output of dynamic statistics is something like this:
647
648 ---------------------------------------------------------------------------------------------------------------------------------
649 Get Statistics
650 Type Time(s) Ops TPS(ops/s) Net(M/s) Get_miss Min(us) Max(us)
651 Avg(us) Std_dev Geo_dist
652 Period 5 345826 69165 65.3 0 27 2198 203
653 95.43 177.29
654 Global 20 1257935 62896 71.8 0 26 3791 224
655 117.79 192.60
656
657
658 Set Statistics
659 Type Time(s) Ops TPS(ops/s) Net(M/s) Get_miss Min(us) Max(us)
660 Avg(us) Std_dev Geo_dist
661 Period 5 38425 7685 7.3 0 42 628 240
662 88.05 220.21
663 Global 20 139780 6989 8.0 0 37 3790 253
664 117.93 224.83
665
666
667 Total Statistics
668 Type Time(s) Ops TPS(ops/s) Net(M/s) Get_miss Min(us) Max(us)
669 Avg(us) Std_dev Geo_dist
670 Period 5 384252 76850 72.5 0 27 2198 207
671 94.72 181.18
672 Global 20 1397720 69886 79.7 0 26 3791 227
673 117.93 195.60
674 ---------------------------------------------------------------------------------------------------------------------------------
675
676 =head2 Where
677
678 =over 4
679
680 =item Get Statistics
681
682 Statistics information of get command
683
684 =item Set Statistics
685
686 Statistics information of set command
687
688 =item Total Statistics
689
690 Statistics information of both get and set command
691
692 =item Period
693
694 Result within a period
695
696 =item Global
697
698 Accumulated results
699
700 =item Ops
701
702 Total operations
703
704 =item TPS
705
706 Throughput, operations/second
707
708 =item Net
709
710 The rate of network
711
712 =item Get_miss
713
714 How many objects can’t be gotten
715
716 =item Min
717
718 The minimum response time
719
720 =item Max
721
722 The maximum response time
723
724 =item Avg:
725
726 The average response time
727
728 =item Std_dev
729
730 Standard deviation of response time
731
732 =item Geo_dist
733
734 Geometric distribution based on natural exponential function
735
736 =back
737
738 At the end, memslap will output something like this:
739
740 ---------------------------------------------------------------------------------------------------------------------------------
741 Get Statistics (1257956 events)
742 Min: 26
743 Max: 3791
744 Avg: 224
745 Geo: 192.60
746 Std: 116.23
747 Log2 Dist:
748 4: 0 10 84490 215345
749 8: 484890 459823 12543 824
750 12: 31
751
752 Set Statistics (139782 events)
753 Min: 37
754 Max: 3790
755 Avg: 253
756 Geo: 224.84
757 Std: 116.83
758 Log2 Dist:
759 4: 0 0 4200 16988
760 8: 50784 65574 2064 167
761 12: 5
762
763 Total Statistics (1397738 events)
764 Min: 26
765 Max: 3791
766 Avg: 227
767 Geo: 195.60
768 Std: 116.60
769 Log2 Dist:
770 4: 0 10 88690 232333
771 8: 535674 525397 14607 991
772 12: 36
773
774 cmd_get: 1257969
775 cmd_set: 139785
776 get_misses: 0
777 verify_misses: 0
778 verify_failed: 0
779 expired_get: 0
780 unexpired_unget: 0
781 written_bytes: 242516030
782 read_bytes: 1003702556
783 object_bytes: 152086080
784 packet_disorder: 0
785 packet_drop: 0
786 udp_timeout: 0
787
788 Run time: 20.0s Ops: 1397754 TPS: 69817 Net_rate: 59.4M/s
789 ---------------------------------------------------------------------------------------------------------------------------------
790
791 =head2 Where
792
793 =over 4
794
795 =item Get Statistics
796
797 Get statistics of response time
798
799 =item Set Statistics
800
801 Set statistics of response time
802
803 =item Total Statistics
804
805 Both get and set statistics of response time
806
807 =item Min
808
809 The accumulated and minimum response time
810
811 =item Max
812
813 The accumulated and maximum response time
814
815 =item Avg
816
817 The accumulated and average response time
818
819 =item Std
820
821 Standard deviation of response time
822
823 =item Log2 Dist
824
825 Geometric distribution based on logarithm 2
826
827 =item cmd_get
828
829 Total get commands done
830
831 =item cmd_set
832
833 Total set commands done
834
835 =item get_misses
836
837 How many objects can’t be gotten from server
838
839 =item verify_misses
840
841 How many objects need to verify but can’t get them
842
843 =item verify_failed
844
845 How many objects with insistent value
846
847 =item expired_get
848
849 How many objects are expired but we get them
850
851 =item unexpired_unget
852
853 How many objects are unexpired but we can’t get them
854
855 =item written_bytes
856
857 Total written bytes
858
859 =item read_bytes
860
861 Total read bytes
862
863 =item object_bytes
864
865 Total object bytes
866
867 =item packet_disorder
868
869 How many UDP packages are disorder
870
871 =item packet_drop
872
873 How many UDP packages are lost
874
875 =item udp_timeout
876
877 How many times UDP time out happen
878
879 =item Run time
880
881 Total run time
882
883 =item Ops
884
885 Total operations
886
887 =item TPS
888
889 Throughput, operations/second
890
891 =item Net_rate
892
893 The average rate of network
894
895 =back
896
897 =head1 OPTIONS
898
899 -s, --servers=
900 List one or more servers to connect. Servers count must be less than
901 threads count. e.g.: --servers=localhost:1234,localhost:11211
902
903 -T, --threads=
904 Number of threads to startup, better equal to CPU numbers. Default 8.
905
906 -c, --concurrency=
907 Number of concurrency to simulate with load. Default 128.
908
909 -n, --conn_sock=
910 Number of TCP socks per concurrency. Default 1.
911
912 -x, --execute_number=
913 Number of operations(get and set) to execute for the
914 given test. Default 1000000.
915
916 -t, --time=
917 How long the test to run, suffix: s-seconds, m-minutes, h-hours,
918 d-days e.g.: --time=2h.
919
920 -F, --cfg_cmd=
921 Load the configure file to get command,key and value distribution list.
922
923 -w, --win_size=
924 Task window size of each concurrency, suffix: K, M e.g.: --win_size=10k.
925 Default 10k.
926
927 -X, --fixed_size=
928 Fixed length of value.
929
930 -v, --verify=
931 The proportion of date verification, e.g.: --verify=0.01
932
933 -d, --division=
934 Number of keys to multi-get once. Default 1, means single get.
935
936 -S, --stat_freq=
937 Frequency of dumping statistic information. suffix: s-seconds,
938 m-minutes, e.g.: --resp_freq=10s.
939
940 -e, --exp_verify=
941 The proportion of objects with expire time, e.g.: --exp_verify=0.01.
942 Default no object with expire time
943
944 -o, --overwrite=
945 The proportion of objects need overwrite, e.g.: --overwrite=0.01.
946 Default never overwrite object.
947
948 -R, --reconnect
949 Reconnect support, when connection is closed it will be reconnected.
950
951 -U, --udp
952 UDP support, default memslap uses TCP, TCP port and UDP port of
953 server must be same.
954
955 -a, --facebook
956 Whether it enables facebook test feature, set with TCP and multi-get with UDP.
957
958 -B, --binary
959 Whether it enables binary protocol. Default with ASCII protocol.
960
961 -P, --tps=
962 Expected throughput, suffix: K, e.g.: --tps=10k.
963
964 -p, --rep_write=
965 The first nth servers can write data, e.g.: --rep_write=2.
966
967 -b, --verbose
968 Whether it outputs detailed information when verification fails.
969
970 -h, --help
971 Display this message and then exit.
972
973 -V, --version
974 Display the version of the application and then exit.
975
976 =head1 EXAMPLES
977
978 memslap -s 127.0.0.1:11211 -S 5s
979
980 memslap -s 127.0.0.1:11211 -t 2m -v 0.2 -e 0.05 -b
981
982 memslap -s 127.0.0.1:11211 -F config -t 2m -w 40k -S 20s -o 0.2
983
984 memslap -s 127.0.0.1:11211 -F config -t 2m -T 4 -c 128 -d 20 -P 40k
985
986 memslap -s 127.0.0.1:11211 -F config -t 2m -d 50 -a -n 40
987
988 memslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m
989
990 memslap -s 127.0.0.1:11211,127.0.0.1:11212 -F config -t 2m -p 2
991
992 =head1 HOME
993
994 To find out more information please check:
995 L<http://launchpad.org/libmemcached>
996
997 =head1 AUTHORS
998
999 Mingqiang Zhuang E<lt>mingqiangzhuang@hengtiansoft.comE<gt> (Schooner Technolgy)
1000 Brian Aker, E<lt>brian@tangent.orgE<gt>
1001
1002 =head1 SEE ALSO
1003
1004 memcached(1) libmemcached(3)
1005
1006 =cut
1007