f9dac0525765da9f86628dec5448c5d3e9df10c8
[m6w6/libmemcached] / docs / man / memaslap.1
1 .TH "MEMASLAP" "1" "April 08, 2011" "0.47" "libmemcached"
2 .SH NAME
3 memaslap \- libmemcached Documentation
4 .
5 .nr rst2man-indent-level 0
6 .
7 .de1 rstReportMargin
8 \\$1 \\n[an-margin]
9 level \\n[rst2man-indent-level]
10 level margin: \\n[rst2man-indent\\n[rst2man-indent-level]]
11 -
12 \\n[rst2man-indent0]
13 \\n[rst2man-indent1]
14 \\n[rst2man-indent2]
15 ..
16 .de1 INDENT
17 .\" .rstReportMargin pre:
18 . RS \\$1
19 . nr rst2man-indent\\n[rst2man-indent-level] \\n[an-margin]
20 . nr rst2man-indent-level +1
21 .\" .rstReportMargin post:
22 ..
23 .de UNINDENT
24 . RE
25 .\" indent \\n[an-margin]
26 .\" old: \\n[rst2man-indent\\n[rst2man-indent-level]]
27 .nr rst2man-indent-level -1
28 .\" new: \\n[rst2man-indent\\n[rst2man-indent-level]]
29 .in \\n[rst2man-indent\\n[rst2man-indent-level]]u
30 ..
31 .\" Man page generated from reStructeredText.
32 .
33 .SH SYNOPSIS
34 .INDENT 0.0
35 .TP
36 .B \-\-help
37 .UNINDENT
38 .SH DESCRIPTION
39 .sp
40 \fBmemaslap\fP is a load generation and benchmark tool for memcached(1)
41 servers. It generates configurable workload such as threads, concurrencies, connections,
42 run time, overwrite, miss rate, key size, value size, get/set proportion,
43 expected throughput, and so on. Furthermore, it also testss data
44 verification, expire\-time verification, UDP, binary protocol, facebook test,
45 replication test, multi\-get and reconnection, etc.
46 .sp
47 Memslap manages network connections like memcached with
48 libevent. Each thread of memaslap is bound with a CPU core, all
49 the threads don\(aqt communicate with each other, and there are several socket
50 connections in each thread. Each connection keeps key size distribution,
51 value size distribution, and command distribution by itself.
52 .sp
53 You can specify servers via the \fB\-\-servers\fP option or via the
54 environment variable \fBMEMCACHED_SERVERS\fP.
55 .SH FEATURES
56 .sp
57 Memslap is developed to for the following purposes:
58 .sp
59 Manages network connections with libevent asynchronously.
60 .sp
61 Set both TCP and UDP up to use non\-blocking IO.
62 .sp
63 Improves parallelism: higher performance in multi\-threads environments.
64 .sp
65 Improves time efficiency: faster processing speed.
66 .sp
67 Generates key and value more efficiently; key size distribution and value size distribution are configurable.
68 .sp
69 Supports get, multi\-get, and set commands; command distribution is configurable.
70 .sp
71 Supports controllable miss rate and overwrite rate.
72 .sp
73 Supports data and expire\-time verification.
74 .sp
75 Supports dumping statistic information periodically.
76 .sp
77 Supports thousands of TCP connections.
78 .sp
79 Supports binary protocol.
80 .sp
81 Supports facebook test (set with TCP and multi\-get with UDP) and replication test.
82 .SH DETAILS
83 .SS Effective implementation of network.
84 .sp
85 For memaslap, both TCP and UDP use non\-blocking network IO. All
86 the network events are managed by libevent as memcached. The network module
87 of memaslap is similar to memcached. Libevent can ensure
88 memaslap can handle network very efficiently.
89 .SS Effective implementation of multi\-threads and concurrency
90 .sp
91 Memslap has the similar implementation of multi\-threads to
92 memcached. Memslap creates one or more self\-governed threads;
93 each thread is bound with one CPU core if the system testss setting CPU
94 core affinity.
95 .sp
96 In addition, each thread has a libevent to manage the events of the network;
97 each thread has one or more self\-governed concurrencies; and each
98 concurrency has one or more socket connections. All the concurrencies don’t
99 communicate with each other even though they are in the same thread.
100 .sp
101 Memslap can create thousands of socket connections, and each
102 concurrency has tens of socket connections. Each concurrency randomly or
103 sequentially selects one socket connection from its socket connection pool
104 to run, so memaslap can ensure each concurrency handles one
105 socket connection at any given time. Users can specify the number of
106 concurrency and socket connections of each concurrency according to their
107 expected workload.
108 .SS Effective implementation of generating key and value
109 .sp
110 In order to improve time efficiency and space efficiency,
111 memaslap creates a random characters table with 10M characters. All the
112 suffixes of keys and values are generated from this random characters table.
113 .sp
114 Memslap uses the offset in the character table and the length
115 of the string to identify a string. It can save much memory.
116 Each key contains two parts, a prefix and a suffix. The prefix is an
117 uint64_t, 8 bytes. In order to verify the data set before,
118 memaslap need to ensure each key is unique, so it uses the prefix to identify
119 a key. The prefix cannot include illegal characters, such as ‘r’, ‘n’,
120 ‘0’ and ‘ ‘. And memaslap has an algorithm to ensure that.
121 .sp
122 Memslap doesn’t generate all the objects (key\-value pairs) at
123 the beginning. It only generates enough objects to fill the task window
124 (default 10K objects) of each concurrency. Each object has the following
125 basic information, key prefix, key suffix offset in the character table, key
126 length, value offset in the character table, and value length.
127 .sp
128 In the work process, each concurrency sequentially or randomly selects an
129 object from the window to do set operation or get operation. At the same
130 time, each concurrency kicks objects out of its window and adds new object
131 into it.
132 .SS Simple but useful task scheduling
133 .sp
134 Memslap uses libevent to schedule all the concurrencies of
135 threads, and each concurrency schedules tasks based on the local task
136 window. Memslap assumes that if each concurrency keeps the same
137 key distribution, value distribution and commands distribution, from
138 outside, memaslap keeps all the distribution as a whole.
139 Each task window includes a lot of objects, each object stores its basic
140 information, such as key, value, expire time, and so on. At any time, all
141 the objects in the window keep the same and fixed key and value
142 distribution. If an object is overwritten, the value of the object will be
143 updated. Memslap verifies the data or expire\-time according to
144 the object information stored in the task window.
145 .sp
146 Libevent selects which concurrency to handle based on a specific network
147 event. Then the concurrency selects which command (get or set) to operate
148 based on the command distribution. If it needs to kick out an old object and
149 add a new object, in order to keep the same key and value distribution, the
150 new object must have the same key length and value length.
151 .sp
152 If memcached server has two cache layers (memory and SSD), running
153 memaslap with different window sizes can get different cache
154 miss rates. If memaslap adds enough objects into the windows at
155 the beginning, and the cache of memcached cannot store all the objects
156 initialized, then memaslap will get some objects from the second
157 cache layer. It causes the first cache layer to miss. So the user can
158 specify the window size to get the expected miss rate of the first cache
159 layer.
160 .SS Useful implementation of multi\-servers , UDP, TCP, multi\-get and binary protocol
161 .sp
162 Because each thread is self\-governed, memaslap can assign
163 different threads to handle different memcached servers. This is just one of
164 the ways in which memaslap testss multiple servers. The only
165 limitation is that the number of servers cannot be greater than the number
166 of threads. The other way to tests multiple servers is for replication
167 test. Each concurrency has one socket connection to each memcached server.
168 For the implementation, memaslap can set some objects to one
169 memcached server, and get these objects from the other servers.
170 .sp
171 By default, Memslap does single get. If the user specifies
172 multi\-get option, memaslap will collect enough get commands and
173 pack and send the commands together.
174 .sp
175 Memslap testss both the ASCII protocol and binary protocol,
176 but it runs on the ASCII protocol by default.
177 Memslap by default runs on the TCP protocol, but it also
178 testss UDP. Because UDP is unreliable, dropped packages and out\-of\-order
179 packages may occur. Memslap creates a memory buffer to handle
180 these problems. Memslap tries to read all the response data of
181 one command from the server and reorders the response data. If some packages
182 get lost, the waiting timeout mechanism can ensure half\-baked packages will
183 be discarded and the next command will be sent.
184 .SH USAGE
185 .sp
186 Below are some usage samples:
187 .sp
188 memaslap \-s 127.0.0.1:11211 \-S 5s
189 .sp
190 memaslap \-s 127.0.0.1:11211 \-t 2m \-v 0.2 \-e 0.05 \-b
191 .sp
192 memaslap \-s 127.0.0.1:11211 \-F config \-t 2m \-w 40k \-S 20s \-o 0.2
193 .sp
194 memaslap \-s 127.0.0.1:11211 \-F config \-t 2m \-T 4 \-c 128 \-d 20 \-P 40k
195 .sp
196 memaslap \-s 127.0.0.1:11211 \-F config \-t 2m \-d 50 \-a \-n 40
197 .sp
198 memaslap \-s 127.0.0.1:11211,127.0.0.1:11212 \-F config \-t 2m
199 .sp
200 memaslap \-s 127.0.0.1:11211,127.0.0.1:11212 \-F config \-t 2m \-p 2
201 .sp
202 The user must specify one server at least to run memaslap. The
203 rest of the parameters have default values, as shown below:
204 .sp
205 Thread number = 1 Concurrency = 16
206 .sp
207 Run time = 600 seconds Configuration file = NULL
208 .sp
209 Key size = 64 Value size = 1024
210 .sp
211 Get/set = 9:1 Window size = 10k
212 .sp
213 Execute number = 0 Single get = true
214 .sp
215 Multi\-get = false Number of sockets of each concurrency = 1
216 .sp
217 Reconnect = false Data verification = false
218 .sp
219 Expire\-time verification = false ASCII protocol = true
220 .sp
221 Binary protocol = false Dumping statistic information
222 .sp
223 periodically = false
224 .sp
225 Overwrite proportion = 0% UDP = false
226 .sp
227 TCP = true Limit throughput = false
228 .sp
229 Facebook test = false Replication test = false
230 .SS Key size, value size and command distribution.
231 .sp
232 All the distributions are read from the configuration file specified by user
233 with “—cfg_cmd” option. If the user does not specify a configuration file,
234 memaslap will run with the default distribution (key size = 64,
235 value size = 1024, get/set = 9:1). For information on how to edit the
236 configuration file, refer to the “Configuration File” section.
237 .sp
238 The minimum key size is 16 bytes; the maximum key size is 250 bytes. The
239 precision of proportion is 0.001. The proportion of distribution will be
240 rounded to 3 decimal places.
241 .sp
242 The minimum value size is 1 bytes; the maximum value size is 1M bytes. The
243 precision of proportion is 0.001. The proportion of distribution will be
244 rounded to 3 decimal places.
245 Currently, memaslap only testss set and get commands. And it
246 testss 100% set and 100% get. For 100% get, it will preset some objects to
247 the server.
248 .SS Multi\-thread and concurrency
249 .sp
250 The high performance of memaslap benefits from the special
251 schedule of thread and concurrency. It’s important to specify the proper
252 number of them. The default number of threads is 1; the default number of
253 concurrency is 16. The user can use “—threads” and “\-\-concurrency” to
254 specify these variables.
255 .sp
256 If the system testss setting CPU affinity and the number of threads
257 specified by the user is greater than 1, memaslap will try to
258 bind each thread to a different CPU core. So if you want to get the best
259 performance memaslap, it is better to specify the number of
260 thread equal to the number of CPU cores. The number of threads specified by
261 the user can also be less or greater than the number of CPU cores. Because
262 of the limitation of implementation, the number of concurrencies could be
263 the multiple of the number of threads.
264 .INDENT 0.0
265 .IP 1. 3
266 .
267 For 8 CPU cores system
268 .UNINDENT
269 .sp
270 For example:
271 .sp
272 \-\-threads=2 \-\-concurrency=128
273 .sp
274 \-\-threads=8 \-\-concurrency=128
275 .sp
276 \-\-threads=8 \-\-concurrency=256
277 .sp
278 \-\-threads=12 \-\-concurrency=144
279 .INDENT 0.0
280 .IP 2. 3
281 .
282 For 16 CPU cores system
283 .UNINDENT
284 .sp
285 For example:
286 .sp
287 \-\-threads=8 \-\-concurrency=128
288 .sp
289 \-\-threads=16 \-\-concurrency=256
290 .sp
291 \-\-threads=16 \-\-concurrency=512
292 .sp
293 \-\-threads=24 \-\-concurrency=288
294 .sp
295 The memaslap performs very well, when
296 used to test the performance of memcached servers.
297 Most of the time, the bottleneck is the network or
298 the server. If for some reason the user wants to
299 limit the performance of memaslap, there
300 are two ways to do this:
301 .sp
302 Decrease the number of threads and concurrencies.
303 Use the option “\-\-tps” that memaslap
304 provides to limit the throughput. This option allows
305 the user to get the expected throughput. For
306 example, assume that the maximum throughput is 50
307 kops/s for a specific configuration, you can specify
308 the throughput equal to or less than the maximum
309 throughput using “\-\-tps” option.
310 .SS Window size
311 .sp
312 Most of the time, the user does not need to specify the window size. The
313 default window size is 10k. For Schooner Memcached, the user can specify
314 different window sizes to get different cache miss rates based on the test
315 case. Memslap testss cache miss rate between 0% and 100%.
316 If you use this utility to test the performance of Schooner Memcached, you
317 can specify a proper window size to get the expected cache miss rate. The
318 formula for calculating window size is as follows:
319 .sp
320 Assume that the key size is 128 bytes, and the value size is 2048 bytes, and
321 concurrency=128.
322 .sp
323 1. Small cache cache_size=1M, 100% cache miss (all data get from SSD).
324 win_size=10k
325 .INDENT 0.0
326 .IP 2. 3
327 .
328 cache_size=4G
329 .UNINDENT
330 .sp
331 (1). cache miss rate 0%
332 .sp
333 win_size=8k
334 .sp
335 (2). cache miss rate 5%
336 .sp
337 win_size=11k
338 .INDENT 0.0
339 .IP 3. 3
340 .
341 cache_size=16G
342 .UNINDENT
343 .sp
344 (1). cache miss rate 0%
345 .sp
346 win_size=32k
347 .sp
348 (2). cache miss
349 .sp
350 rate 5%
351 .sp
352 win_size=46k
353 .sp
354 The formula for calculating window size for cache miss rate 0%:
355 .sp
356 cache_size / concurrency / (key_size + value_size) * 0.5
357 .sp
358 The formula for calculating window size for cache miss rate 5%:
359 .sp
360 cache_size / concurrency / (key_size + value_size) * 0.7
361 .SS Verification
362 .sp
363 Memslap testss both data verification and expire\-time
364 verification. The user can use "\-\-verify=" or "\-v" to specify the proportion
365 of data verification. In theory, it testss 100% data verification. The
366 user can use "\-\-exp_verify=" or "\-e" to specify the proportion of
367 expire\-time verification. In theory, it testss 100% expire\-time
368 verification. Specify the "\-\-verbose" options to get more detailed error
369 information.
370 .sp
371 For example: \-\-exp_verify=0.01 –verify=0.1 , it means that 1% of the objects
372 set with expire\-time, 10% of the objects gotten will be verified. If the
373 objects are gotten, memaslap will verify the expire\-time and
374 value.
375 .SS multi\-servers and multi\-config
376 .sp
377 Memslap testss multi\-servers based on self\-governed thread.
378 There is a limitation that the number of servers cannot be greater than the
379 number of threads. Memslap assigns one thread to handle one
380 server at least. The user can use the "\-\-servers=" or "\-s" option to specify
381 multi\-servers.
382 .sp
383 For example:
384 .sp
385 \-\-servers=10.1.1.1:11211,10.1.1.2:11212,10.1.1.3:11213 \-\-threads=6 \-\-concurrency=36
386 .sp
387 The above command means that there are 6 threads, with each thread having 6
388 concurrencies and that threads 0 and 3 handle server 0 (10.1.1.1); threads 1
389 and 4 handle server 1 (10.1.1.2); and thread 2 and 5 handle server 2
390 (10.1.1.3).
391 .sp
392 All the threads and concurrencies in memaslap are self\-governed.
393 .sp
394 So is memaslap. The user can start up several
395 memaslap instances. The user can run memaslap on different client
396 machines to communicate with the same memcached server at the same. It is
397 recommended that the user start different memaslap on different
398 machines using the same configuration.
399 .SS Run with execute number mode or time mode
400 .sp
401 The default memaslap runs with time mode. The default run time
402 is 10 minutes. If it times out, memaslap will exit. Do not
403 specify both execute number mode and time mode at the same time; just
404 specify one instead.
405 .sp
406 For example:
407 .sp
408 \-\-time=30s (It means the test will run 30 seconds.)
409 .sp
410 \-\-execute_number=100000 (It means that after running 100000 commands, the test will exit.)
411 .SS Dump statistic information periodically.
412 .sp
413 The user can use "\-\-stat_freq=" or "\-S" to specify the frequency.
414 .sp
415 For example:
416 .sp
417 \-\-stat_freq=20s
418 .sp
419 Memslap will dump the statistics of the commands (get and set) at the frequency of every 20
420 seconds.
421 .sp
422 For more information on the format of dumping statistic information, refer to “Format of Output” section.
423 .SS Multi\-get
424 .sp
425 The user can use "\-\-division=" or "\-d" to specify multi\-get keys count.
426 Memslap by default does single get with TCP. Memslap also testss data
427 verification and expire\-time verification for multi\-get.
428 .sp
429 Memslap testss multi\-get with both TCP and UDP. Because of
430 the different implementation of the ASCII protocol and binary protocol,
431 there are some differences between the two. For the ASCII protocol,
432 memaslap sends one “multi\-get” to the server once. For the
433 binary protocol, memaslap sends several single get commands
434 together as “multi\-get” to the server.
435 .SS UDP and TCP
436 .sp
437 Memslap testss both UDP and TCP. For TCP,
438 memaslap does not reconnect the memcached server if socket connections are
439 lost. If all the socket connections are lost or memcached server crashes,
440 memaslap will exit. If the user specifies the “\-\-reconnect”
441 option when socket connections are lost, it will reconnect them.
442 .sp
443 User can use “\-\-udp” to enable the UDP feature, but UDP comes with some
444 limitations:
445 .sp
446 UDP cannot set data more than 1400 bytes.
447 .sp
448 UDP is not testsed by the binary protocol because the binary protocol of
449 memcached does not tests that.
450 .sp
451 UDP doesn’t tests reconnection.
452 .SS Facebook test
453 .sp
454 Set data with TCP and multi\-get with UDP. Specify the following options:
455 .sp
456 "\-\-facebook \-\-division=50"
457 .sp
458 If you want to create thousands of TCP connections, specify the
459 .sp
460 "\-\-conn_sock=" option.
461 .sp
462 For example: \-\-facebook \-\-division=50 \-\-conn_sock=200
463 .sp
464 The above command means that memaslap will do facebook test,
465 each concurrency has 200 socket TCP connections and one UDP socket.
466 .sp
467 Memslap sets objects with the TCP socket, and multi\-gets 50
468 objects once with the UDP socket.
469 .sp
470 If you specify "\-\-division=50", the key size must be less that 25 bytes
471 because the UDP packet size is 1400 bytes.
472 .SS Replication test
473 .sp
474 For replication test, the user must specify at least two memcached servers.
475 The user can use “—rep_write=” option to enable feature.
476 .sp
477 For example:
478 .sp
479 \-\-servers=10.1.1.1:11211,10.1.1.2:11212 –rep_write=2
480 .sp
481 The above command means that there are 2 replication memcached servers,
482 memaslap will set objects to both server 0 and server 1, get
483 objects which are set to server 0 before from server 1, and also get objects
484 which are set to server 1 before from server 0. If server 0 crashes,
485 memaslap will only get objects from server 1. If server 0 comes
486 back to life again, memaslap will reconnect server 0. If both
487 server 0 and server 1 crash, memaslap will exit.
488 .SS Supports thousands of TCP connections
489 .sp
490 Start memaslap with "\-\-conn_sock=" or "\-n" to enable this
491 feature. Make sure that your system can tests opening thousands of files
492 and creating thousands of sockets. However, this feature does not tests
493 reconnection if sockets disconnect.
494 .sp
495 For example:
496 .sp
497 \-\-threads=8 \-\-concurrency=128 \-\-conn_sock=128
498 .sp
499 The above command means that memaslap starts up 8 threads, each
500 thread has 16 concurrencies, each concurrency has 128 TCP socket
501 connections, and the total number of TCP socket connections is 128 * 128 =
502 16384.
503 .SS Supports binary protocol
504 .sp
505 Start memaslap with "\-\-binary" or "\-B" options to enable this
506 feature. It testss all the above features except UDP, because the latest
507 memcached 1.3.3 does not implement binary UDP protocol.
508 .sp
509 For example:
510 .sp
511 \-\-binary
512 .sp
513 Since memcached 1.3.3 doesn\(aqt implement binary UDP protocol,
514 memaslap does not tests UDP. In addition, memcached 1.3.3 does not tests
515 multi\-get. If you specify "\-\-division=50" option, it just sends 50 get
516 commands together as “mulit\-get” to the server.
517 .SH CONFIGURATION FILE
518 .sp
519 This section describes the format of the configuration file. By default
520 when no configuration file is specified memaslap reads the default
521 one located at ~/.memaslap.cnf.
522 .sp
523 Below is a sample configuration file:
524 .sp
525 .nf
526 .ft C
527 \-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-
528 #comments should start with \(aq#\(aq
529 #key
530 #start_len end_len proportion
531 #
532 #key length range from start_len to end_len
533 #start_len must be equal to or greater than 16
534 #end_len must be equal to or less than 250
535 #start_len must be equal to or greater than end_len
536 #memaslap will generate keys according to the key range
537 #proportion: indicates keys generated from one range accounts for the total
538 generated keys
539 #
540 #example1: key range 16~100 accounts for 80%
541 # key range 101~200 accounts for 10%
542 # key range 201~250 accounts for 10%
543 # total should be 1 (0.8+0.1+0.1 = 1)
544 #
545 # 16 100 0.8
546 # 101 200 0.1
547 # 201 249 0.1
548 #
549 #example2: all keys length are 128 bytes
550 #
551 # 128 128 1
552 key
553 128 128 1
554 #value
555 #start_len end_len proportion
556 #
557 #value length range from start_len to end_len
558 #start_len must be equal to or greater than 1
559 #end_len must be equal to or less than 1M
560 #start_len must be equal to or greater than end_len
561 #memaslap will generate values according to the value range
562 #proportion: indicates values generated from one range accounts for the
563 total generated values
564 #
565 #example1: value range 1~1000 accounts for 80%
566 # value range 1001~10000 accounts for 10%
567 # value range 10001~100000 accounts for 10%
568 # total should be 1 (0.8+0.1+0.1 = 1)
569 #
570 # 1 1000 0.8
571 # 1001 10000 0.1
572 # 10001 100000 0.1
573 #
574 #example2: all value length are 128 bytes
575 #
576 # 128 128 1
577 value
578 2048 2048 1
579 #cmd
580 #cmd_type cmd_proportion
581 #
582 #currently memaslap only testss get and set command.
583 #
584 #cmd_type
585 #set 0
586 #get 1
587 #
588 #example: set command accounts for 50%
589 # get command accounts for 50%
590 # total should be 1 (0.5+0.5 = 1)
591 #
592 # cmd
593 # 0 0.5
594 # 1 0.5
595 cmd
596 0 0.1
597 1.0 0.9
598 .ft P
599 .fi
600 .SH FORMAT OF OUTPUT
601 .sp
602 At the beginning, memaslap displays some configuration information as follows:
603 .sp
604 servers : 127.0.0.1:11211
605 .sp
606 threads count: 1
607 .sp
608 concurrency: 16
609 .sp
610 run time: 20s
611 .sp
612 windows size: 10k
613 .sp
614 set proportion: set_prop=0.10
615 .sp
616 get proportion: get_prop=0.90
617 .SS Where
618 .sp
619 servers : "servers"
620 .INDENT 0.0
621 .INDENT 3.5
622 .sp
623 The servers used by memaslap.
624 .UNINDENT
625 .UNINDENT
626 .sp
627 threads count
628 .INDENT 0.0
629 .INDENT 3.5
630 .sp
631 The number of threads memaslap runs with.
632 .UNINDENT
633 .UNINDENT
634 .sp
635 concurrency
636 .INDENT 0.0
637 .INDENT 3.5
638 .sp
639 The number of concurrencies memaslap runs with.
640 .UNINDENT
641 .UNINDENT
642 .sp
643 run time
644 .INDENT 0.0
645 .INDENT 3.5
646 .sp
647 How long to run memaslap.
648 .UNINDENT
649 .UNINDENT
650 .sp
651 windows size
652 .INDENT 0.0
653 .INDENT 3.5
654 .sp
655 The task window size of each concurrency.
656 .UNINDENT
657 .UNINDENT
658 .sp
659 set proportion
660 .INDENT 0.0
661 .INDENT 3.5
662 .sp
663 The proportion of set command.
664 .UNINDENT
665 .UNINDENT
666 .sp
667 get proportion
668 .INDENT 0.0
669 .INDENT 3.5
670 .sp
671 The proportion of get command.
672 .UNINDENT
673 .UNINDENT
674 .sp
675 The output of dynamic statistics is something like this:
676 .sp
677 .nf
678 .ft C
679 \-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-
680 Get Statistics
681 Type Time(s) Ops TPS(ops/s) Net(M/s) Get_miss Min(us) Max(us)
682 Avg(us) Std_dev Geo_dist
683 Period 5 345826 69165 65.3 0 27 2198 203
684 95.43 177.29
685 Global 20 1257935 62896 71.8 0 26 3791 224
686 117.79 192.60
687
688
689 Set Statistics
690 Type Time(s) Ops TPS(ops/s) Net(M/s) Get_miss Min(us) Max(us)
691 Avg(us) Std_dev Geo_dist
692 Period 5 38425 7685 7.3 0 42 628 240
693 88.05 220.21
694 Global 20 139780 6989 8.0 0 37 3790 253
695 117.93 224.83
696
697
698 Total Statistics
699 Type Time(s) Ops TPS(ops/s) Net(M/s) Get_miss Min(us) Max(us)
700 Avg(us) Std_dev Geo_dist
701 Period 5 384252 76850 72.5 0 27 2198 207
702 94.72 181.18
703 Global 20 1397720 69886 79.7 0 26 3791 227
704 117.93 195.60
705 \-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-
706 .ft P
707 .fi
708 .SS Where
709 .sp
710 Get Statistics
711 .INDENT 0.0
712 .INDENT 3.5
713 .sp
714 Statistics information of get command
715 .UNINDENT
716 .UNINDENT
717 .sp
718 Set Statistics
719 .INDENT 0.0
720 .INDENT 3.5
721 .sp
722 Statistics information of set command
723 .UNINDENT
724 .UNINDENT
725 .sp
726 Total Statistics
727 .INDENT 0.0
728 .INDENT 3.5
729 .sp
730 Statistics information of both get and set command
731 .UNINDENT
732 .UNINDENT
733 .sp
734 Period
735 .INDENT 0.0
736 .INDENT 3.5
737 .sp
738 Result within a period
739 .UNINDENT
740 .UNINDENT
741 .sp
742 Global
743 .INDENT 0.0
744 .INDENT 3.5
745 .sp
746 Accumulated results
747 .UNINDENT
748 .UNINDENT
749 .sp
750 Ops
751 .INDENT 0.0
752 .INDENT 3.5
753 .sp
754 Total operations
755 .UNINDENT
756 .UNINDENT
757 .sp
758 TPS
759 .INDENT 0.0
760 .INDENT 3.5
761 .sp
762 Throughput, operations/second
763 .UNINDENT
764 .UNINDENT
765 .sp
766 Net
767 .INDENT 0.0
768 .INDENT 3.5
769 .sp
770 The rate of network
771 .UNINDENT
772 .UNINDENT
773 .sp
774 Get_miss
775 .INDENT 0.0
776 .INDENT 3.5
777 .sp
778 How many objects can’t be gotten
779 .UNINDENT
780 .UNINDENT
781 .sp
782 Min
783 .INDENT 0.0
784 .INDENT 3.5
785 .sp
786 The minimum response time
787 .UNINDENT
788 .UNINDENT
789 .sp
790 Max
791 .INDENT 0.0
792 .INDENT 3.5
793 .sp
794 The maximum response time
795 .UNINDENT
796 .UNINDENT
797 .sp
798 Avg:
799 .INDENT 0.0
800 .INDENT 3.5
801 .sp
802 The average response time
803 .UNINDENT
804 .UNINDENT
805 .sp
806 Std_dev
807 .INDENT 0.0
808 .INDENT 3.5
809 .sp
810 Standard deviation of response time
811 .UNINDENT
812 .UNINDENT
813 .sp
814 Geo_dist
815 .INDENT 0.0
816 .INDENT 3.5
817 .sp
818 Geometric distribution based on natural exponential function
819 .UNINDENT
820 .UNINDENT
821 .sp
822 At the end, memaslap will output something like this:
823 .sp
824 .nf
825 .ft C
826 \-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-
827 Get Statistics (1257956 events)
828 Min: 26
829 Max: 3791
830 Avg: 224
831 Geo: 192.60
832 Std: 116.23
833 Log2 Dist:
834 4: 0 10 84490 215345
835 8: 484890 459823 12543 824
836 12: 31
837
838 Set Statistics (139782 events)
839 Min: 37
840 Max: 3790
841 Avg: 253
842 Geo: 224.84
843 Std: 116.83
844 Log2 Dist:
845 4: 0 0 4200 16988
846 8: 50784 65574 2064 167
847 12: 5
848
849 Total Statistics (1397738 events)
850 Min: 26
851 Max: 3791
852 Avg: 227
853 Geo: 195.60
854 Std: 116.60
855 Log2 Dist:
856 4: 0 10 88690 232333
857 8: 535674 525397 14607 991
858 12: 36
859
860 cmd_get: 1257969
861 cmd_set: 139785
862 get_misses: 0
863 verify_misses: 0
864 verify_failed: 0
865 expired_get: 0
866 unexpired_unget: 0
867 written_bytes: 242516030
868 read_bytes: 1003702556
869 object_bytes: 152086080
870 packet_disorder: 0
871 packet_drop: 0
872 udp_timeout: 0
873
874 Run time: 20.0s Ops: 1397754 TPS: 69817 Net_rate: 59.4M/s
875 \-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-
876 .ft P
877 .fi
878 .SS Where
879 .sp
880 Get Statistics
881 .INDENT 0.0
882 .INDENT 3.5
883 .sp
884 Get statistics of response time
885 .UNINDENT
886 .UNINDENT
887 .sp
888 Set Statistics
889 .INDENT 0.0
890 .INDENT 3.5
891 .sp
892 Set statistics of response time
893 .UNINDENT
894 .UNINDENT
895 .sp
896 Total Statistics
897 .INDENT 0.0
898 .INDENT 3.5
899 .sp
900 Both get and set statistics of response time
901 .UNINDENT
902 .UNINDENT
903 .sp
904 Min
905 .INDENT 0.0
906 .INDENT 3.5
907 .sp
908 The accumulated and minimum response time
909 .UNINDENT
910 .UNINDENT
911 .sp
912 Max
913 .INDENT 0.0
914 .INDENT 3.5
915 .sp
916 The accumulated and maximum response time
917 .UNINDENT
918 .UNINDENT
919 .sp
920 Avg
921 .INDENT 0.0
922 .INDENT 3.5
923 .sp
924 The accumulated and average response time
925 .UNINDENT
926 .UNINDENT
927 .sp
928 Std
929 .INDENT 0.0
930 .INDENT 3.5
931 .sp
932 Standard deviation of response time
933 .UNINDENT
934 .UNINDENT
935 .sp
936 Log2 Dist
937 .INDENT 0.0
938 .INDENT 3.5
939 .sp
940 Geometric distribution based on logarithm 2
941 .UNINDENT
942 .UNINDENT
943 .sp
944 cmd_get
945 .INDENT 0.0
946 .INDENT 3.5
947 .sp
948 Total get commands done
949 .UNINDENT
950 .UNINDENT
951 .sp
952 cmd_set
953 .INDENT 0.0
954 .INDENT 3.5
955 .sp
956 Total set commands done
957 .UNINDENT
958 .UNINDENT
959 .sp
960 get_misses
961 .INDENT 0.0
962 .INDENT 3.5
963 .sp
964 How many objects can’t be gotten from server
965 .UNINDENT
966 .UNINDENT
967 .sp
968 verify_misses
969 .INDENT 0.0
970 .INDENT 3.5
971 .sp
972 How many objects need to verify but can’t get them
973 .UNINDENT
974 .UNINDENT
975 .sp
976 verify_failed
977 .INDENT 0.0
978 .INDENT 3.5
979 .sp
980 How many objects with insistent value
981 .UNINDENT
982 .UNINDENT
983 .sp
984 expired_get
985 .INDENT 0.0
986 .INDENT 3.5
987 .sp
988 How many objects are expired but we get them
989 .UNINDENT
990 .UNINDENT
991 .sp
992 unexpired_unget
993 .INDENT 0.0
994 .INDENT 3.5
995 .sp
996 How many objects are unexpired but we can’t get them
997 .UNINDENT
998 .UNINDENT
999 .sp
1000 written_bytes
1001 .INDENT 0.0
1002 .INDENT 3.5
1003 .sp
1004 Total written bytes
1005 .UNINDENT
1006 .UNINDENT
1007 .sp
1008 read_bytes
1009 .INDENT 0.0
1010 .INDENT 3.5
1011 .sp
1012 Total read bytes
1013 .UNINDENT
1014 .UNINDENT
1015 .sp
1016 object_bytes
1017 .INDENT 0.0
1018 .INDENT 3.5
1019 .sp
1020 Total object bytes
1021 .UNINDENT
1022 .UNINDENT
1023 .sp
1024 packet_disorder
1025 .INDENT 0.0
1026 .INDENT 3.5
1027 .sp
1028 How many UDP packages are disorder
1029 .UNINDENT
1030 .UNINDENT
1031 .sp
1032 packet_drop
1033 .INDENT 0.0
1034 .INDENT 3.5
1035 .sp
1036 How many UDP packages are lost
1037 .UNINDENT
1038 .UNINDENT
1039 .sp
1040 udp_timeout
1041 .INDENT 0.0
1042 .INDENT 3.5
1043 .sp
1044 How many times UDP time out happen
1045 .UNINDENT
1046 .UNINDENT
1047 .sp
1048 Run time
1049 .INDENT 0.0
1050 .INDENT 3.5
1051 .sp
1052 Total run time
1053 .UNINDENT
1054 .UNINDENT
1055 .sp
1056 Ops
1057 .INDENT 0.0
1058 .INDENT 3.5
1059 .sp
1060 Total operations
1061 .UNINDENT
1062 .UNINDENT
1063 .sp
1064 TPS
1065 .INDENT 0.0
1066 .INDENT 3.5
1067 .sp
1068 Throughput, operations/second
1069 .UNINDENT
1070 .UNINDENT
1071 .sp
1072 Net_rate
1073 .INDENT 0.0
1074 .INDENT 3.5
1075 .sp
1076 The average rate of network
1077 .UNINDENT
1078 .UNINDENT
1079 .SH OPTIONS
1080 .INDENT 0.0
1081 .TP
1082 .B \-s, \-\-servers=
1083 .
1084 List one or more servers to connect. Servers count must be less than
1085 threads count. e.g.: \-\-servers=localhost:1234,localhost:11211
1086 .TP
1087 .B \-T, \-\-threads=
1088 .
1089 Number of threads to startup, better equal to CPU numbers. Default 8.
1090 .TP
1091 .B \-c, \-\-concurrency=
1092 .
1093 Number of concurrency to simulate with load. Default 128.
1094 .TP
1095 .B \-n, \-\-conn_sock=
1096 .
1097 Number of TCP socks per concurrency. Default 1.
1098 .TP
1099 .B \-x, \-\-execute_number=
1100 .
1101 Number of operations(get and set) to execute for the
1102 given test. Default 1000000.
1103 .TP
1104 .B \-t, \-\-time=
1105 .
1106 How long the test to run, suffix: s\-seconds, m\-minutes, h\-hours,
1107 d\-days e.g.: \-\-time=2h.
1108 .TP
1109 .B \-F, \-\-cfg_cmd=
1110 .
1111 Load the configure file to get command,key and value distribution list.
1112 .TP
1113 .B \-w, \-\-win_size=
1114 .
1115 Task window size of each concurrency, suffix: K, M e.g.: \-\-win_size=10k.
1116 Default 10k.
1117 .TP
1118 .B \-X, \-\-fixed_size=
1119 .
1120 Fixed length of value.
1121 .TP
1122 .B \-v, \-\-verify=
1123 .
1124 The proportion of date verification, e.g.: \-\-verify=0.01
1125 .TP
1126 .B \-d, \-\-division=
1127 .
1128 Number of keys to multi\-get once. Default 1, means single get.
1129 .TP
1130 .B \-S, \-\-stat_freq=
1131 .
1132 Frequency of dumping statistic information. suffix: s\-seconds,
1133 m\-minutes, e.g.: \-\-resp_freq=10s.
1134 .TP
1135 .B \-e, \-\-exp_verify=
1136 .
1137 The proportion of objects with expire time, e.g.: \-\-exp_verify=0.01.
1138 Default no object with expire time
1139 .TP
1140 .B \-o, \-\-overwrite=
1141 .
1142 The proportion of objects need overwrite, e.g.: \-\-overwrite=0.01.
1143 Default never overwrite object.
1144 .UNINDENT
1145 .INDENT 0.0
1146 .TP
1147 .B \-R, \-\-reconnect
1148 .
1149 Reconnect tests, when connection is closed it will be reconnected.
1150 .TP
1151 .B \-U, \-\-udp
1152 .
1153 UDP tests, default memaslap uses TCP, TCP port and UDP port of
1154 server must be same.
1155 .TP
1156 .B \-a, \-\-facebook
1157 .
1158 Whether it enables facebook test feature, set with TCP and multi\-get with UDP.
1159 .TP
1160 .B \-B, \-\-binary
1161 .
1162 Whether it enables binary protocol. Default with ASCII protocol.
1163 .UNINDENT
1164 .INDENT 0.0
1165 .TP
1166 .B \-P, \-\-tps=
1167 .
1168 Expected throughput, suffix: K, e.g.: \-\-tps=10k.
1169 .TP
1170 .B \-p, \-\-rep_write=
1171 .
1172 The first nth servers can write data, e.g.: \-\-rep_write=2.
1173 .UNINDENT
1174 .INDENT 0.0
1175 .TP
1176 .B \-b, \-\-verbose
1177 .
1178 Whether it outputs detailed information when verification fails.
1179 .TP
1180 .B \-h, \-\-help
1181 .
1182 Display this message and then exit.
1183 .TP
1184 .B \-V, \-\-version
1185 .
1186 Display the version of the application and then exit.
1187 .UNINDENT
1188 .SH EXAMPLES
1189 .sp
1190 memaslap \-s 127.0.0.1:11211 \-S 5s
1191 .sp
1192 memaslap \-s 127.0.0.1:11211 \-t 2m \-v 0.2 \-e 0.05 \-b
1193 .sp
1194 memaslap \-s 127.0.0.1:11211 \-F config \-t 2m \-w 40k \-S 20s \-o 0.2
1195 .sp
1196 memaslap \-s 127.0.0.1:11211 \-F config \-t 2m \-T 4 \-c 128 \-d 20 \-P 40k
1197 .sp
1198 memaslap \-s 127.0.0.1:11211 \-F config \-t 2m \-d 50 \-a \-n 40
1199 .sp
1200 memaslap \-s 127.0.0.1:11211,127.0.0.1:11212 \-F config \-t 2m
1201 .sp
1202 memaslap \-s 127.0.0.1:11211,127.0.0.1:11212 \-F config \-t 2m \-p 2
1203 .SH HOME
1204 .sp
1205 To find out more information please check:
1206 \fI\%http://launchpad.org/libmemcached\fP
1207 .SH AUTHORS
1208 .sp
1209 Mingqiang Zhuang <\fI\%mingqiangzhuang@hengtiansoft.com\fP> (Schooner Technolgy)
1210 Brian Aker, <\fI\%brian@tangent.org\fP>
1211 .SH SEE ALSO
1212 .sp
1213 \fImemcached(1)\fP \fIlibmemcached(3)\fP
1214 .SH AUTHOR
1215 Brian Aker
1216 .SH COPYRIGHT
1217 2011, Brian Aker DataDifferential, http://datadifferential.com/
1218 .\" Generated by docutils manpage writer.
1219 .\"
1220 .