Redis-benchmark

명령 목록

• 기본 Benchmark   기본 Benchmark를 한다.
• 클러스터 지원   클러스터를 Benchmark를 한다.
• 단순하게 표시   -q 옵션을 사용한다.
• CSV 형식으로 표시   --csv
• 명령 지정   -t 옵션으로 benchmark를 실행할 명령을 지정한다.
• 실행 횟수 지정   -n 옵션으로 실행 횟수를 지정한다. 기본은 100,000이다.
• Client 수 지정   -c 옵션으로 클라이언트 수를 지정한다. 기본은 50이다.
• Sleep 지정   -sleep 옵션으로 sleep를 지정한다. 단위는 us(microsecond)이다. 기본은 0이다.
• Thread 수 지정   --threads 옵션으로 쓰레드 수를 지정한다.
• Value 사이즈 지정   -d 옵션으로 value 크기를 지정할 수 있다. 기본은 3바이트이다.
• Key, Value 다양하게 지정   -r 옵션으로 key 또는 value를 다양하게 테스트할 수 있다.
• Key, Value 직접 지정   -r 옵션과 같이 사용한다.
• Client tracking on 지정   Client side caching을 참조하세요.
• 측정 정밀도 지정   -precision   0|1|2
• HELP   도움말을 보여준다.

기본 Benchmark

• 주요 명령을 10만회씩 실행해서 성능을 측정한다. 주요 명령은 PINK, SET, GET, INCR, LPUSH, RPUSH, LPOP, RPOP, SADD, HSET, SPOP, LRANGE, MSET 이다.
• 테스트 서버에서 주요 입력 명령의 초당 처리량(실행 횟수)은 약 20만회이다. $ src/redis-benchmark -p 6001 ====== PING_INLINE ====== 100000 requests completed in 0.95 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.92% <= 1 milliseconds 100.00% <= 1 milliseconds 104821.80 requests per second ====== PING_BULK ====== 100000 requests completed in 0.55 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 181488.20 requests per second ====== SET ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 199203.20 requests per second ====== GET ====== 100000 requests completed in 0.51 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 197628.47 requests per second ====== INCR ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 198807.16 requests per second ====== LPUSH ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 200400.80 requests per second ====== RPUSH ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 200803.22 requests per second ====== LPOP ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 200000.00 requests per second ====== RPOP ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 198019.80 requests per second ====== SADD ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 199600.80 requests per second ====== HSET ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 199600.80 requests per second ====== SPOP ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 198807.16 requests per second ====== LPUSH (needed to benchmark LRANGE) ====== 100000 requests completed in 0.50 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 200803.22 requests per second ====== LRANGE_100 (first 100 elements) ====== 100000 requests completed in 1.16 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 86058.52 requests per second ====== LRANGE_300 (first 300 elements) ====== 100000 requests completed in 2.91 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.96% <= 1 milliseconds 100.00% <= 1 milliseconds 34387.89 requests per second ====== LRANGE_500 (first 450 elements) ====== 100000 requests completed in 4.96 seconds 50 parallel clients 3 bytes payload keep alive: 1 26.60% <= 1 milliseconds 99.99% <= 2 milliseconds 100.00% <= 2 milliseconds 20149.10 requests per second ====== LRANGE_600 (first 600 elements) ====== 100000 requests completed in 5.24 seconds 50 parallel clients 3 bytes payload keep alive: 1 0.05% <= 1 milliseconds 99.96% <= 2 milliseconds 99.99% <= 3 milliseconds 100.00% <= 3 milliseconds 19069.41 requests per second ====== MSET (10 keys) ====== 100000 requests completed in 0.71 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 141643.06 requests per second
• 버전 6.0부터 save, appendonly, threads 가 추가로 표시된다. 이 항목(옵션)에 따라 성능이 달라질 수 있다. $ src/redis-benchmark -p 6001 -t set -d 100 --threads 8 ====== SET ====== 100000 requests completed in 3.52 seconds 50 parallel clients 100 bytes payload keep alive: 1 host configuration "save": host configuration "appendonly": yes multi-thread: yes threads: 8

클러스터 지원: --cluster

• 버전 6.0부터 클러스터에 benchmark가 가능하다. [charlie@CentOS7 redis-6.0-rc1]$ src/redis-benchmark -p 7000 -t set -d 100 --cluster Cluster has 3 master nodes: Master 0: 3dc49ccc0dddeb8a621b28744ddee82cf14af040 127.0.0.1:7001 Master 1: 6960872da1aa36e8c4670adbafa3c47f4001efbc 127.0.0.1:7002 Master 2: 745e6d7d18e2501b196d401aa8160bb37b15f754 127.0.0.1:7000 ====== SET ====== 100000 requests completed in 3.26 seconds 50 parallel clients 100 bytes payload keep alive: 1 cluster mode: yes (3 masters) node [0] configuration: save: NONE appendonly: no node [1] configuration: save: NONE appendonly: no node [2] configuration: save: NONE appendonly: no multi-thread: yes threads: 3
• 버전 6.0 미만에서 클러스터로 구성된 하나의 노드에 benchmark를 실행하면 benchmark는 실행되는 것으로 표시되지만 cluster redis-server에서는 처리되지 않는다. 즉, SET, GET 같은 명령이 실행되지 않는다. 그러므로 benchmark 실행 후 해당 노드에 접속해서 데이터를 확인하면 하나도 없고 당연히 aof 파일에도 저장되지 않는다. 다만 ping 명령은 실행된다. Ping 명령으로도 성능을 가늠해볼 수는 있다.
• 버전 6.0 미만에서는 단독 마스터를 실행해서 benchmark를 테스트한다.

단순하게 표시: -q

• $ src/redis-benchmark -p 6001 -q PING_INLINE: 166666.66 requests per second PING_BULK: 199203.20 requests per second SET: 201612.91 requests per second GET: 199203.20 requests per second INCR: 201612.91 requests per second LPUSH: 202429.16 requests per second RPUSH: 202429.16 requests per second LPOP: 200400.80 requests per second RPOP: 201612.91 requests per second SADD: 200803.22 requests per second HSET: 202429.16 requests per second SPOP: 200803.22 requests per second LPUSH (needed to benchmark LRANGE): 202839.75 requests per second LRANGE_100 (first 100 elements): 86430.43 requests per second LRANGE_300 (first 300 elements): 34328.87 requests per second LRANGE_500 (first 450 elements): 24752.47 requests per second LRANGE_600 (first 600 elements): 18986.14 requests per second MSET (10 keys): 139860.14 requests per second

CSV 형식으로 표시: --csv

• $ src/redis-benchmark -p 6001 --csv "PING_INLINE","169204.73" "PING_BULK","196463.66" "SET","199600.80" "GET","197628.47" "INCR","198807.16" "LPUSH","199600.80" "RPUSH","198807.16" "LPOP","198807.16" "RPOP","198019.80" "SADD","198807.16" "HSET","200400.80" "SPOP","200400.80" "LPUSH (needed to benchmark LRANGE)","200000.00" "LRANGE_100 (first 100 elements)","85984.52" "LRANGE_300 (first 300 elements)","34211.43" "LRANGE_500 (first 450 elements)","24491.79" "LRANGE_600 (first 600 elements)","18957.35" "MSET (10 keys)","140646.97"

명령 지정: -t

• -t 옵션으로 benchmark를 실행할 명령을 지정한다. $ src/redis-benchmark -p 6001 -t set ====== SET ====== 100000 requests completed in 0.58 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 171232.88 requests per second

실행 횟수 지정: -n

• -n 옵션으로 실행 횟수를 지정한다. Default는 100,000이다. $ src/redis-benchmark -p 6001 -t set -n 500000 ====== SET ====== 500000 requests completed in 2.61 seconds 50 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 191570.89 requests per second

Client 수 지정: -c

• -c 옵션으로 Clients 수를 지정한다. Default는 50이다. $ src/redis-benchmark -p 6001 -t set -n 500000 -c 10 ====== SET ====== 500000 requests completed in 2.56 seconds 10 parallel clients 3 bytes payload keep alive: 1 100.00% <= 0 milliseconds 195694.72 requests per second

Sleep 지정: -usleep

• -usleep로 sleep를 지정한다. 단위는 마이크로초(microsecond)이다. 1000000 -> 1초, 1000 -> 1ms 입니다. 위에서 설명한 client 옵션과 같이 사용해서 실행 명령 수를 조정할 수 있습니다. Default는 0이다.
  이 옵션은 Enterprise 7.2.0부터 사용할 수 있습니다. $ src/redis-benchmark -p 6001 -t set -n 500000 -usleep 100

Thread 수 지정: --threads

• --threads <num> 옵션으로 쓰레드 수를 지정한다. Default는 No이다. $ src/redis-benchmark -p 6001 -t set -n 500000 --threads 8 ====== SET ====== 500000 requests completed in 17.04 seconds 50 parallel clients 3 bytes payload keep alive: 1 host configuration "save": host configuration "appendonly": yes multi-thread: yes threads: 8

Value 사이즈 지정: -d

• -d 옵션으로 Value 사이즈를 지정한다. Default는 3이다. Payload에 지정한 사이즈(바이트)가 표시된다.
• SET 명령의 경우 키 이름이 "key:__rand_int__"이다. $ src/redis-benchmark -p 6001 -t set -d 20 ====== SET ====== 100000 requests completed in 0.59 seconds 50 parallel clients 20 bytes payload keep alive: 1 100.00% <= 0 milliseconds 168918.92 requests per second $ src/redis-cli -p 6001 127.0.0.1:6001> get key:__rand_int__ "xxxxxxxxxxxxxxxxxxxx" $ src/redis-benchmark -p 6001 -t lpush -d 20 ====== LPUSH ====== 100000 requests completed in 0.59 seconds 50 parallel clients 20 bytes payload keep alive: 1 100.00% <= 0 milliseconds 170357.75 requests per second 127.0.0.1:6001> lrange mylist 0 5 1) "xxxxxxxxxxxxxxxxxxxx" 2) "xxxxxxxxxxxxxxxxxxxx" 3) "xxxxxxxxxxxxxxxxxxxx" 4) "xxxxxxxxxxxxxxxxxxxx" 5) "xxxxxxxxxxxxxxxxxxxx" 6) "xxxxxxxxxxxxxxxxxxxx"

Key, Value 다양하게 지정: -r

• -r 옵션으로 key 또는 value를 다양하게 테스트할 수 있다.
• SET test $ src/redis-benchmark -p 6001 -t set -r 100000 ====== SET ====== 100000 requests completed in 0.95 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.87% <= 1 milliseconds 100.00% <= 1 milliseconds 104931.80 requests per second $ src/redis-cli -p 6001 127.0.0.1:6001> dbsize (integer) 63251 127.0.0.1:6001> scan 0 1) "53248" 2) 1) "key:000000080588" 2) "key:000000027240" 3) "key:000000009027" 4) "key:000000028279" 5) "key:000000057324" 6) "key:000000044128" 7) "key:000000035869" 8) "key:000000027983" 9) "key:000000003973" 10) "key:000000002511"
• SADD test $ src/redis-benchmark -p 6001 -t sadd -n 100 -r 100000 ====== SADD ====== 100 requests completed in 0.00 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.00% <= 1 milliseconds 100.00% <= 1 milliseconds 50000.00 requests per second $ src/redis-cli -p 6001 127.0.0.1:6001> scard myset (integer) 100 127.0.0.1:6001> sscan myset 0 1) "24" 2) 1) "element:000000037570" 2) "element:000000004318" 3) "element:000000030866" 4) "element:000000052834" 5) "element:000000065944" 6) "element:000000012470" 7) "element:000000018875" 8) "element:000000061924" 9) "element:000000007416" 10) "element:000000084806"

Key, Value 직접 지정

• LPUSH: redis-benchmark -p 6001 -r 1000000 lpush mylist ele:__rand_int__ $ src/redis-benchmark -p 6001 -r 1000000 lpush mylist ele:__rand_int__ ====== lpush mylist ele:__rand_int__ ====== 100000 requests completed in 0.94 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.88% <= 1 milliseconds 100.00% <= 1 milliseconds 105820.11 requests per second $ src/redis-cli -p 6001 127.0.0.1:6001> llen mylist (integer) 100000 127.0.0.1:6001> lrange mylist 0 6 1) "ele:000000423107" 2) "ele:000000346547" 3) "ele:000000216990" 4) "ele:000000280357" 5) "ele:000000762985" 6) "ele:000000956186"
• SADD: redis-benchmark -p 6001 -r 1000000 sadd myset mem:__rand_int__
• redis-benchmark -p 6001 -n 1000000 -r 100000000 set keyA___rand_int__ valueA___rand_int__
• redis-benchmark -p 6001 -n 1000000 -r 100000000 get keyA___rand_int__
• redis-benchmark -p 6001 -n 1000000 -r 100000000 del keyA___rand_int__
• redis-benchmark -p 6001 -n 1000000 -r 100000000 zadd myzset 0 memA___rand_int__
• redis-benchmark -p 6001 -n 1000000 get keyA_000000000345
• redis-benchmark -p 6001 -n 1000000 object time keyA_000000000345
• redis-benchmark -p 6001 -n 1000000 rpop mylist5


• -r 숫자를 크게하면 member 중복을 적게 할 수 있다. $ src/redis-benchmark -p 6001 -r 1000000 sadd myset mem:__rand_int__ ====== sadd myset mem:__rand_int__ ====== 100000 requests completed in 0.60 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.90% <= 1 milliseconds 100.00% <= 1 milliseconds 166944.92 requests per second $ src/redis-cli -p 6001 127.0.0.1:6001> scard myset (integer) 95177 127.0.0.1:6001> sscan myset 0 1) "12288" 2) 1) "mem:000000170342" 2) "mem:000000803395" 3) "mem:000000301154" 4) "mem:000000949710" 5) "mem:000000489267" 6) "mem:000000542854" 7) "mem:000000767008" 8) "mem:000000451243" 9) "mem:000000638887" 10) "mem:000000724675" 11) "mem:000000970089" 12) "mem:000000575203"
• ZADD: redis-benchmark -p 6001 -r 1000000 zadd myzset __rand_int__ member:__rand_int__ $ src/redis-benchmark -p 6001 -r 1000000 zadd myzset __rand_int__ member:__rand_int__ ====== zadd myzset __rand_int__ member:__rand_int__ ====== 100000 requests completed in 0.64 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.94% <= 1 milliseconds 100.00% <= 1 milliseconds 156739.81 requests per second $ src/redis-cli -p 6001 127.0.0.1:6001> zcard myzset (integer) 95122 127.0.0.1:6001> zscan myzset 0 1) "90112" 2) 1) "member:000000433048" 2) "105823" 3) "member:000000650524" 4) "345104" 5) "member:000000034870" 6) "779946" 7) "member:000000622620" 8) "130322"
• HSET: redis-benchmark -p 6001 -r 1000000 hset myhash field:__rand_int__ value:__rand_int__ $ src/redis-benchmark -p 6001 -r 1000000 hset myhash field:__rand_int__ value:__rand_int__ ====== hset myhash field:__rand_int__ value:__rand_int__ ====== 100000 requests completed in 0.94 seconds 50 parallel clients 3 bytes payload keep alive: 1 99.36% <= 1 milliseconds 99.96% <= 2 milliseconds 100.00% <= 2 milliseconds 106044.54 requests per second $ src/redis-cli -p 6001 127.0.0.1:6001> hscan myhash 0 1) "40960" 2) 1) "field:000000878442" 2) "value:000000620020" 3) "field:000000168038" 4) "value:000000141778" 5) "field:000000815313" 6) "value:000000852776"

Client tracking on: --enable-tracking

• $ src/redis-benchmark -p 6001 -t set -d 100 --enable-tracking

측정(표시) 정밀도: --precision

• --precision 0: millisecond 단위로 표시한다.
  $ src/redis-benchmark -p 6001 -t set -d 100 --precision 0
  27.88% <= 1 milliseconds
  98.78% <= 2 milliseconds
  99.76% <= 3 milliseconds
• --precision 1: 0.1 millisecond 단위로 표시한다.
  $ src/redis-benchmark -p 6001 -t set -d 100 --precision 1
  0.12% <= 0.6 milliseconds
  0.32% <= 0.7 milliseconds
  5.91% <= 0.8 milliseconds
• --precision 2: 0.01 millisecond 단위로 표시한다.
  $ src/redis-benchmark -p 6001 -t set -d 100 --precision 2
  0.03% <= 0.68 milliseconds
  0.12% <= 0.69 milliseconds
  0.28% <= 0.70 milliseconds

HELP

• $ src/redis-benchmark --help Usage: redis-benchmark [-h <host>] [-p <port>] [-c <clients>] [-n <requests>] [-k <boolean>] -h <hostname> Server hostname (default 127.0.0.1) -p <port> Server port (default 6379) -s <socket> Server socket (overrides host and port) -a <password> Password for Redis Auth -c <clients> Number of parallel connections (default 50) -n <requests> Total number of requests (default 100000) -d <size> Data size of SET/GET value in bytes (default 3) --dbnum <db> SELECT the specified db number (default 0) --threads <num> Enable multi-thread mode. ver 6.0 --cluster Enable cluster mode. ver 6.0 --enable-tracking Send CLIENT TRACKING on before starting benchmark. ver 6.0 -k <boolean> 1=keep alive 0=reconnect (default 1) -r <keyspacelen> Use random keys for SET/GET/INCR, random values for SADD Using this option the benchmark will expand the string __rand_int__ inside an argument with a 12 digits number in the specified range from 0 to keyspacelen-1. The substitution changes every time a command is executed. Default tests use this to hit random keys in the specified range. -P <numreq> Pipeline <numreq> requests. Default 1 (no pipeline). -e If server replies with errors, show them on stdout. (no more than 1 error per second is displayed) -q Quiet. Just show query/sec values --precision Number of decimal places to display in latency output (default 0) ver 6.0 --csv Output in CSV format -l Loop. Run the tests forever -t <tests> Only run the comma separated list of tests. The test names are the same as the ones produced as output. -I Idle mode. Just open N idle connections and wait. Examples: Run the benchmark with the default configuration against 127.0.0.1:6379: $ redis-benchmark Use 20 parallel clients, for a total of 100k requests, against 192.168.1.1: $ redis-benchmark -h 192.168.1.1 -p 6379 -n 100000 -c 20 Fill 127.0.0.1:6379 with about 1 million keys only using the SET test: $ redis-benchmark -t set -n 1000000 -r 100000000 Benchmark 127.0.0.1:6379 for a few commands producing CSV output: $ redis-benchmark -t ping,set,get -n 100000 --csv Benchmark a specific command line: $ redis-benchmark -r 10000 -n 10000 eval 'return redis.call("ping")' 0 Fill a list with 10000 random elements: $ redis-benchmark -r 10000 -n 10000 lpush mylist __rand_int__ On user specified command lines __rand_int__ is replaced with a random integer with a range of values selected by the -r option.