A quick comparison of spring boot per-request resource usage with reactive vs traditional MVC methods.
This experiment is contrived - the difference between the three examples below is just the threading behaviour at the web request handling layer. Your results will differ if you introduce more variation in underlying IO operations, but in general the following results are shown:
- Traditional MVC used the most threads, Coroutines about half as much, and Reactor even less.
- Traditional MVC used the most heap/memory, Coroutines less, and Reactor even less.
Some thoughts following from this:
- If minimising cost of computing per request is your only goal then choose Reactor, at the expense of imperative coding styles.
- Coroutines give definite memory & threading advantages over MVC without changing coding styles.
- Install some java resource monitoring tool, e.g.:
sdk install visualvm - Start up the monitoring tool.
- Start a backing database, e.g.:
docker run -e POSTGRES_USER=postgres -e POSTGRES_PASSWORD=postgres -e POSTGRES_DB=db -p 5432:5432 -d postgres:16 - Build all subprojects:
.gradlew assemble
- Run:
java -jar ./traditional/build/libs/traditional-0.0.1-SNAPSHOT.jar - Attach the monitoring tool to the JVM process.
- Run:
java -jar ./kotlin-coroutines/build/libs/kotlin-coroutines-0.0.1-SNAPSHOT.jar - Attach the monitoring tool to the JVM process.
- Run:
java -jar ./project-reactor/build/libs/project-reactor-0.0.1-SNAPSHOT.jar - Attach the monitoring tool to the JVM process.
- Start load testing, e.g.:
ab -n 20000 -c 50 -s 60 http://localhost:8080/api/nextvalab -n 20000 -c 50 -s 60 http://localhost:8081/api/nextvalab -n 20000 -c 50 -s 60 http://localhost:8082/api/nextval
- Results, e.g.:
ab -n 20000 -c 50 -s 60 http://localhost:8080/api/nextval
This is ApacheBench, Version 2.3 <$Revision: 1903618 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 2000 requests
Completed 4000 requests
Completed 6000 requests
Completed 8000 requests
Completed 10000 requests
Completed 12000 requests
Completed 14000 requests
Completed 16000 requests
Completed 18000 requests
Completed 20000 requests
Finished 20000 requests
Server Software:
Server Hostname: localhost
Server Port: 8080
Document Path: /api/nextval
Document Length: 17 bytes
Concurrency Level: 50
Time taken for tests: 37.765 seconds
Complete requests: 20000
Failed requests: 2762
(Connect: 0, Receive: 0, Length: 2762, Exceptions: 0)
Total transferred: 2442762 bytes
HTML transferred: 342762 bytes
Requests per second: 529.59 [#/sec] (mean)
Time per request: 94.413 [ms] (mean)
Time per request: 1.888 [ms] (mean, across all concurrent requests)
Transfer rate: 63.17 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 88 1748.4 0 35013
Processing: 0 6 10.5 4 246
Waiting: 0 6 10.1 3 244
Total: 1 94 1749.8 4 35053
Percentage of the requests served within a certain time (ms)
50% 4
66% 5
75% 7
80% 8
90% 12
95% 17
98% 28
99% 41
100% 35053 (longest request)
ab -n 20000 -c 50 -s 60 http://localhost:8081/api/nextval
This is ApacheBench, Version 2.3 <$Revision: 1903618 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 2000 requests
Completed 4000 requests
Completed 6000 requests
Completed 8000 requests
Completed 10000 requests
Completed 12000 requests
Completed 14000 requests
Completed 16000 requests
Completed 18000 requests
Completed 20000 requests
Finished 20000 requests
Server Software:
Server Hostname: localhost
Server Port: 8081
Document Path: /api/nextval
Document Length: 18 bytes
Concurrency Level: 50
Time taken for tests: 38.459 seconds
Complete requests: 20000
Failed requests: 0
Total transferred: 1780000 bytes
HTML transferred: 360000 bytes
Requests per second: 520.04 [#/sec] (mean)
Time per request: 96.146 [ms] (mean)
Time per request: 1.923 [ms] (mean, across all concurrent requests)
Transfer rate: 45.20 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 88 1748.2 0 35017
Processing: 1 8 6.6 6 334
Waiting: 1 7 6.6 6 331
Total: 1 95 1749.5 6 35050
Percentage of the requests served within a certain time (ms)
50% 6
66% 7
75% 8
80% 9
90% 12
95% 14
98% 20
99% 37
100% 35050 (longest request)
ab -n 20000 -c 50 -s 60 http://localhost:8082/api/nextval
This is ApacheBench, Version 2.3 <$Revision: 1903618 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 2000 requests
Completed 4000 requests
Completed 6000 requests
Completed 8000 requests
Completed 10000 requests
Completed 12000 requests
Completed 14000 requests
Completed 16000 requests
Completed 18000 requests
Completed 20000 requests
Finished 20000 requests
Server Software:
Server Hostname: localhost
Server Port: 8082
Document Path: /api/nextval
Document Length: 18 bytes
Concurrency Level: 50
Time taken for tests: 38.206 seconds
Complete requests: 20000
Failed requests: 0
Total transferred: 1780000 bytes
HTML transferred: 360000 bytes
Requests per second: 523.48 [#/sec] (mean)
Time per request: 95.515 [ms] (mean)
Time per request: 1.910 [ms] (mean, across all concurrent requests)
Transfer rate: 45.50 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 88 1748.3 0 35012
Processing: 1 7 11.2 5 361
Waiting: 0 7 11.2 5 359
Total: 1 95 1749.0 5 35041
Percentage of the requests served within a certain time (ms)
50% 5
66% 6
75% 6
80% 7
90% 10
95% 13
98% 19
99% 35
100% 35041 (longest request)