fastjson_benchmark

1. 测试环境

1.1 各个库的版本

fastjson2 2.0.13
fastjson1 1.0.83
jackson 2.13.4
gson 2.9.1
kryo 5.3.0
hessian 4.0.66

1.2 服务器环境

测试的服务器基于阿里云最新代ECS，包括x64架构的Intel和AMD处理器，ARM架构的AltraMax和阿里云平头哥的倚天处理器。

cs.c7a.xlarge
ecs.c7a.xlarge
ecs.c6r.xlarge
ecs.g8m.xlarge

ecs.c7.xlarge

这个是阿里云当前代标准型ECS，处理器型号 Intel Xeon(Ice Lake) Platinum 8369B，4核，8G内存

ecs.c7a.xlarge

这个是阿里云当前代标准型ECS，处理器型号 AMD EPYC™ Milan 7T83，4核，8G内存

ecs.c6r.xlarge

这个是阿里云上售卖ARM处理器ECS，处理器型号 Ampere Altra / AltraMax，4核，8G内存

ecs.g8m.xlarge

这个是阿里云上售卖ARM处理器，处理器型号 Yitian 710，4核，16G内存。这个是阿里云平头哥的倚天710处理器，需要联系客服才能购买。

1.3 JDK版本

基于Oracle最新版本的Linux x64/aarch64的JDK版本，下载地址 https://www.oracle.com/java/technologies/

oracle-jdk1.8.0_341
oracle-jdk-11.0.16
oracle-jdk-17.0.4
oracle-jdk-18.0.2

1.4 测试代码以及运行方式

代码路径 https://github.com/alibaba/fastjson2/tree/2.0.13/benchmark/src/main/java/com/alibaba/fastjson2/benchmark/eishay
运行方式测试代码方式

git clone https://github.com/alibaba/fastjson2
cd fastjson2
git checkout 2.0.13
mvn clean install -Dmaven.test.skip
~/Install/jdk-1.8.0_341/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay
~/Install/jdk-11.0.16/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay
~/Install/jdk-17.0.4/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay
~/Install/jdk-18.0.2/bin/java -cp ~/git/fastjson2/benchmark/target/fastjson2-benchmarks.jar com.alibaba.fastjson2.benchmark.eishay.Eishay

2. 测试数据&结果分析

原始数据 https://github.com/alibaba/fastjson2/blob/2.0.13/docs/benchmark/benchmark_2.0.13_raw.md

2.1 EishayParseString

这个是最常用的场景，将JSON格式字符串反序列化为Java对象，这个场景在fastjson中的代码如下：

String str = "...";
Bean bean = JSON.parseJSONObject(str, Bean.class);

这个场景是fastjson1最强的场景；

在ecs.c7.xlarge-oracle-jdk1.8.0_341_x64环境中，只相当于只相当于fastjson2的74.34%;jackson则只相当于fastjson2的38.71%; gson是fastjson2的31.75%，也就是说这个最常见的反序列化场景，fastjson2的性能差不多是jackson/gson的三倍。

性能排序如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

      fastjson2 > fastjson1 > jackson > gson
jdk8  100%        74.36%      38.71%    31.75%
jdk11 100%        74.5%       38.34%    34.38%
jdk17 100%        92.12%      36.91%    33.54%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	1360.568	1011.738 (74.36%)	526.676 (38.71%)	432.017 (31.75%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	1228.065	914.894 (74.5%)	470.822 (38.34%)	422.16 (34.38%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	1364.202	1256.679 (92.12%)	503.53 (36.91%)	457.547 (33.54%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	1370.066	1179.882 (86.12%)	492.247 (35.93%)	420.361 (30.68%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	1286.529	1071.109 (83.26%)	525.017 (40.81%)	454.12 (35.3%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	1249.335	940.833 (75.31%)	485.729 (38.88%)	468.326 (37.49%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	1310.954	1293.647 (98.68%)	515.493 (39.32%)	511.211 (39%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	1147.499	1179.527 (102.79%)	496.404 (43.26%)	445.762 (38.85%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	872.976	719.895 (82.46%)	379.914 (43.52%)	310.103 (35.52%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	871.471	736.203 (84.48%)	377.196 (43.28%)	318.83 (36.59%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	879.445	1015.535 (115.47%)	378.534 (43.04%)	312.391 (35.52%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	880.765	993.02 (112.75%)	361.865 (41.09%)	301.743 (34.26%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	914.457	725.44 (79.33%)	367.634 (40.2%)	339.642 (37.14%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	1080.758	808.698 (74.83%)	405.438 (37.51%)	373.572 (34.57%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	1130.264	1028.981 (91.04%)	410.39 (36.31%)	347.428 (30.74%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	990.989	1136.849 (114.72%)	403.869 (40.75%)	362.801 (36.61%)

2.2 EishayParseStringPretty

这个场景是将格式化过的JSON格式字符串反序列化为Java对象，这个场景fastjson1的parse算法不擅长的。这个场景fastjson2表现仍然非常好。这个场景在ecs.c7.xlarge-oracle-jdk1.8.0_341_x64环境下，fastjson1的性能是fastjson2的27.15，jackson的性能是fastjson2的49.89%，gson是fastjson2的40.59%.

这个场景在fastjson中的代码如下：

// 这里输入的是格式化过后的json字符串
String str = "{\n" +
        "\t\"id\":123\n" +
        "}";
Bean bean = JSON.parseJSONObject(str, Bean.class);

性能排序分别如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

      fastjson2 > jackson > gson  > fastjson1
jdk8  100%        49.89%    40.59%  27.15%
jdk11 100%        52.01%    44.73%  25.96%
jdk17 100%        46.86%    45.33%  29.79%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	989.92	268.764 (27.15%)	493.855 (49.89%)	401.828 (40.59%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	929.77	241.376 (25.96%)	483.608 (52.01%)	415.913 (44.73%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	987.743	294.252 (29.79%)	462.903 (46.86%)	447.734 (45.33%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	970.0	284.132 (29.29%)	452.741 (46.67%)	395.039 (40.73%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	992.633	282.002 (28.41%)	480.261 (48.38%)	411.905 (41.5%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	873.323	251.499 (28.8%)	446.893 (51.17%)	420.166 (48.11%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	891.968	299.75 (33.61%)	488.85 (54.81%)	482.449 (54.09%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	868.572	293.284 (33.77%)	456.26 (52.53%)	410.161 (47.22%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	685.663	208.336 (30.38%)	345.876 (50.44%)	281.817 (41.1%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	668.422	212.383 (31.77%)	344.776 (51.58%)	295.781 (44.25%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	665.231	242.005 (36.38%)	346.526 (52.09%)	294.261 (44.23%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	673.288	228.153 (33.89%)	333.383 (49.52%)	276.64 (41.09%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	725.106	212.969 (29.37%)	339.355 (46.8%)	314.208 (43.33%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	790.702	228.49 (28.9%)	379.777 (48.03%)	336.26 (42.53%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	814.156	259.078 (31.82%)	379.597 (46.62%)	311.307 (38.24%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	798.532	242.843 (30.41%)	356.333 (44.62%)	290.978 (36.44%)

2.3 EishayParseTreeString

这个场景是将JSON格式字符串反序列化为JSONObject，这也是一个最常见的场景之一，这个场景在fastjson中的代码如下：

String str = "...";
JSONObject jsonObject = JSON.parseJSONObject(str);

fastjson2在这个场景也表现了远超fastjson1/jackson/gson的性能。在不同的环境中，fastjson1和jackson都只能相当于fastjson2性能的40%~60%，gson则更差一些。

性能排序如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

      fastjson2 > fastjson1 > jackson  > gson
jdk8  100%        51.72%      49.62%     33.99%
jdk11 100%        47.48%      52.41%     38.34%
jdk17 100%        54.39%      46.57%     28.78%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	997.653	515.969 (51.72%)	495.047 (49.62%)	339.088 (33.99%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	835.382	396.672 (47.48%)	437.821 (52.41%)	320.745 (38.4%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	1088.249	591.931 (54.39%)	506.817 (46.57%)	313.176 (28.78%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	1128.092	563.691 (49.97%)	456.86 (40.5%)	321.169 (28.47%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	1014.7	477.79 (47.09%)	559.775 (55.17%)	374.786 (36.94%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	848.016	423.335 (49.92%)	466.498 (55.01%)	375.204 (44.24%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	980.59	523.15 (53.35%)	554.072 (56.5%)	377.676 (38.52%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	1042.413	527.567 (50.61%)	499.371 (47.91%)	379.75 (36.43%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	561.623	305.922 (54.47%)	359.392 (63.99%)	259.908 (46.28%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	546.837	320.225 (58.56%)	348.703 (63.77%)	262.577 (48.02%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	628.838	358.546 (57.02%)	381.614 (60.69%)	259.51 (41.27%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	632.975	339.814 (53.69%)	364.472 (57.58%)	260.563 (41.16%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	565.736	300.699 (53.15%)	324.266 (57.32%)	262.866 (46.46%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	662.551	361.773 (54.6%)	385.641 (58.21%)	301.285 (45.47%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	827.05	395.672 (47.84%)	424.3 (51.3%)	298.869 (36.14%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	824.21	378.399 (45.91%)	413.141 (50.13%)	280.966 (34.09%)

2.4 EishayParseUTF8Bytes

这个是将UTF8格式的byte数组反序列化为Java对象，这个场景在缓存和RPC场景中常用。

这个场景在fastjson中的代码如下：

byte[] utf8Bytes = ...;
Bean bean = JSON.parseJSONObject(utf8Bytes, Bean.class);

这个场景fastjson2同样表现出了卓越的性能；在JDK8下，fastjson1和jackson的性能分别之后fastjson2的71.23%和53.76%; gson不直接支持输入utf8Bytes，需要先构造字符串，性能只有fastjson2的28.3%。

性能排序如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

      fastjson2 > fastjson1 > jackson  > gson
jdk8  100%        71.23%      53.76%     28.3%
jdk11 100%        68.79%      52.92%     29.55%
jdk17 100%        83.94%      49.69%     26.63%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	1111.437	791.627 (71.23%)	597.483 (53.76%)	314.588 (28.3%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	1067.251	734.136 (68.79%)	565.209 (52.96%)	315.374 (29.55%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	1144.942	961.116 (83.94%)	568.964 (49.69%)	319.794 (27.93%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	1200.863	909.789 (75.76%)	551.602 (45.93%)	319.82 (26.63%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	1048.072	835.454 (79.71%)	616.641 (58.84%)	335.258 (31.99%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	1056.933	715.322 (67.68%)	584.955 (55.34%)	369.708 (34.98%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	1043.35	934.783 (89.59%)	593.304 (56.87%)	382.881 (36.7%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	1067.695	890.046 (83.36%)	583.981 (54.7%)	354.728 (33.22%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	734.367	564.344 (76.85%)	454.68 (61.91%)	239.708 (32.64%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	756.743	590.756 (78.07%)	432.663 (57.17%)	253.007 (33.43%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	784.689	753.819 (96.07%)	433.961 (55.3%)	253.17 (32.26%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	793.883	744.137 (93.73%)	409.848 (51.63%)	251.777 (31.71%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	776.369	651.404 (83.9%)	400.951 (51.64%)	240.136 (30.93%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	981.758	677.353 (68.99%)	470.72 (47.95%)	294.994 (30.05%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	1022.609	604.3 (59.09%)	387.7 (37.91%)	280.935 (27.47%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	1037.552	718.338 (69.23%)	440.787 (42.48%)	280.512 (27.04%)

2.5 EishayWriteString

这个是Java对象序列化成字符串，这个是序列化最常用的场景之一。这个场景在fastjson中的代码如下：

Bean bean = ...;
String str = JSON.toJSONString(bean);

在这个最常用的序列化场景，fastjson2表现出远超fastjson1/jackson/gson的性能；这个场景gson表现很差，只有fastjson2的1/5

性能排序如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

      fastjson2 > jackson >   fastjson1  > gson
jdk8  100%        59.87%      39.28%       20.49%
jdk11 100%        69.88%      43.98%       20.64%
jdk17 100%        61.62%      36.88%       12%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	1576.265	619.203 (39.28%)	943.772 (59.87%)	322.901 (20.49%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	1335.098	587.182 (43.98%)	933.029 (69.88%)	275.623 (20.64%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	1678.255	618.906 (36.88%)	1034.137 (61.62%)	201.428 (12%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	1644.231	593.6 (36.1%)	908.705 (55.27%)	203.305 (12.36%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	1618.597	643.347 (39.75%)	899.484 (55.57%)	376.278 (23.25%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	1440.586	609.974 (42.34%)	950.891 (66.01%)	264.998 (18.4%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	1741.048	640.155 (36.77%)	970.425 (55.74%)	337.577 (19.39%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	1662.442	613.616 (36.91%)	853.506 (51.34%)	318.316 (19.15%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	1044.235	458.396 (43.9%)	571.229 (54.7%)	274.327 (26.27%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	1047.284	427.03 (40.77%)	595.907 (56.9%)	251.026 (23.97%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	1172.453	443.202 (37.8%)	640.451 (54.62%)	189.277 (16.14%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	1125.986	442.262 (39.28%)	584.856 (51.94%)	189.781 (16.85%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	1207.801	488.777 (40.47%)	629.315 (52.1%)	330.416 (27.36%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	1172.745	472.56 (40.3%)	674.348 (57.5%)	288.784 (24.62%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	1347.999	500.066 (37.1%)	605.068 (44.89%)	194.878 (14.46%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	1314.211	448.914 (34.16%)	630.25 (47.96%)	208.63 (15.87%)

2.6 EishayWriteUTF8Bytes

这个是Java对象序列化成UTF格式的byte数组，这个是最常用的场景之一，在缓存和RPC场景常用。gson不直接支持，需要先序列化为String再转换为UTF8格式的byte数组。

这个场景在fastjson中的代码如下：

Bean bean = ...;
byte[] utf8Bytes = JSON.toJSONBytes(bean);

在这个最常用的序列化场景，fastjson2表现出远超fastjson1/jackson/gson的性能；这个场景gson表现很差，只有fastjson2的1/5不到，在JDK 17下差距更大。

性能排序如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

      fastjson2 > jackson >   fastjson1  > gson
jdk8  100%        61.30%      38.15%       19.97%
jdk11 100%        52.29%      33.32%       17.23%
jdk17 100%        51.84%      30.65%       10.82%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2	fastjson1	jackson	gson
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	1483.805	566.016 (38.15%)	909.579 (61.3%)	296.378 (19.97%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	1607.916	535.732 (33.32%)	840.722 (52.29%)	277.027 (17.23%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	1870.324	573.164 (30.65%)	969.668 (51.84%)	202.284 (10.82%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	1842.545	539.979 (29.31%)	922.569 (50.07%)	203.819 (11.06%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	1513.863	577.225 (38.13%)	829.233 (54.78%)	335.696 (22.17%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	1787.631	553.471 (30.96%)	869.786 (48.66%)	252.633 (14.13%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	1703.089	563.399 (33.08%)	866.295 (50.87%)	326.269 (19.16%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	1723.847	535.227 (31.05%)	853.692 (49.52%)	307.143 (17.82%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	1114.874	418.141 (37.51%)	568.404 (50.98%)	235.234 (21.1%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	1155.121	384.802 (33.31%)	553.0 (47.87%)	249.932 (21.64%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	1155.038	386.631 (33.47%)	610.788 (52.88%)	185.482 (16.06%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	1157.346	348.347 (30.1%)	588.702 (50.87%)	185.548 (16.03%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	1322.099	434.992 (32.9%)	644.337 (48.74%)	284.382 (21.51%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	1362.002	424.507 (31.17%)	538.537 (39.54%)	271.529 (19.94%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	1357.4	471.138 (34.71%)	676.224 (49.82%)	202.618 (14.93%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	1421.632	409.363 (28.8%)	630.841 (44.37%)	202.765 (14.26%)

2.7 EishayWriteBinary

这个场景是将Java对象序列化为二进制的byte数组，是缓存序列化和RPC场景使用序列化协议的典型场景。

fastson2内置支持二进制格式jsonb，jsonb是fastjson2为了RPC场景设计的序列化和反序列化设置的高性能序列化协议。

jsonb有两种映射方式，将Java对象映射为类似JSONObject的KV格式，也可以映射类似JSONArray的数组格式。

fastjson2JSONBArrayMapping的代码如下：

Bean bean = ...;
byte[] jsonbBytes = JSONB.toBytes(bean, JSONWriter.Feature.BeanToArray);

fastjson2JSONB的代码如下：

Bean bean = ...;
byte[] jsonbBytes = JSONB.toBytes(bean);

fastjson2UTF8Bytes的代码如下：

Bean bean = ...;
byte[] utf8Bytes = JSON.toJSONBytes(bean);

这个场景下，jsonb的性能非常好，在JDK 11/17下性能更好，hessina和java内置序列化都表现很差。

性能排序如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

      fastjson2JSONB > fastjson2UTF8Bytes> hessian > javaSerialize
jdk8  100%             71.10%              16.14%    11.17%
jdk11 100%             64.15%              13.57%    8.6%
jdk17 100%             55.76%              9.7%      6.69%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2JSONB	fastjson2UTF8Bytes	hessian	javaSerialize
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	2097.46	1491.388 (71.1%)	344.249 (16.41%)	234.23 (11.17%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	2436.506	1562.927 (64.15%)	330.66 (13.57%)	209.629 (8.6%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	3309.755	1845.598 (55.76%)	321.034 (9.7%)	221.56 (6.69%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	3227.077	1828.192 (56.65%)	324.449 (10.05%)	231.471 (7.17%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	2054.803	1504.894 (73.24%)	366.379 (17.83%)	239.777 (11.67%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	2705.693	1764.638 (65.22%)	357.997 (13.23%)	229.602 (8.49%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	2865.521	1596.312 (55.71%)	353.936 (12.35%)	249.249 (8.7%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	2771.114	1749.079 (63.12%)	357.223 (12.89%)	265.558 (9.58%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	1348.365	1103.207 (81.82%)	289.951 (21.5%)	181.14 (13.43%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	1842.171	1145.927 (62.21%)	318.719 (17.3%)	185.214 (10.05%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	1821.557	1155.853 (63.45%)	310.023 (17.02%)	179.866 (9.87%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	1821.794	1159.775 (63.66%)	305.48 (16.77%)	192.008 (10.54%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	1563.48	1300.107 (83.15%)	344.569 (22.04%)	206.703 (13.22%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	2221.289	1323.907 (59.6%)	330.132 (14.86%)	217.229 (9.78%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	2241.298	1309.735 (58.44%)	327.589 (14.62%)	209.455 (9.35%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	2378.319	1427.898 (60.04%)	337.284 (14.18%)	225.171 (9.47%)

2.12 EishayParseBinary

这个场景是将二进制的byte数组反序列化为Java对象，是缓存序列化和RPC场景使用序列化协议的典型场景。

fastjson2JSONBArrayMapping的代码如下：

byte[] jsonbBytes = JSONB.toBytes(bean, JSONWriter.Feature.BeanToArray);
Bean bean = JSONB.parseObject(jsonbBytes, Bean.class, JSONReader.Feature.SupportArrayToBean);

fastjson2JSONB的代码如下：

byte[] jsonbBytes = JSONB.toBytes(bean);
Bean bean = JSONB.parseObject(jsonbBytes, Bean.class);

fastjson2UTF8Bytes的代码如下：

byte[] utf8Bytes = JSON.toJSONBytes(bean);
Bean bean = JSON.parseObject(utf8Bytes, Bean.class);

性能排序如下：

// ecs.c7.xlarge-oracle-jdk1.8.0_341_x64

这个场景下，jsonb的性能非常好，在JDK 11/17下性能更好，hessina表现很差，java内置序列化都表现极差。

      fastjson2JSONB > fastjson2UTF8Bytes> hessian > javaSerialize
jdk8  100%             66.91%              16.85%    2.82%
jdk11 100%             46.97%              10.19%    1.97%
jdk17 100%             42.88%              8.81%     1.85%

下面是不同CPU不同JDK下的性能比较表格，第1列是阿里云机器规格，第2列是JDK的版本信息，第3列到第6列是不同库的JMH的Throughput数据，越大越好。第4列到第6列中的百分比是Throughput相比fastjson2的比，比如第三列的百分比是fastjson1/fastjson2的百分数表示。

aliyun ecs spec	jdk version	fastjson2JSONB	fastjson2UTF8Bytes	hessian	javaSerialize
ecs.c7.xlarge	oracle-jdk1.8.0_341_x64	1649.494	1103.643 (66.91%)	277.963 (16.85%)	46.566 (2.82%)
ecs.c7.xlarge	oracle-jdk-11.0.16_x64	2438.929	1143.151 (46.87%)	248.418 (10.19%)	48.052 (1.97%)
ecs.c7.xlarge	oracle-jdk-17.0.4_x64	2939.697	1260.484 (42.88%)	258.96 (8.81%)	54.268 (1.85%)
ecs.c7.xlarge	oracle-jdk-18.0.2_x64	2836.159	1272.521 (44.87%)	244.107 (8.61%)	51.249 (1.81%)
ecs.c7a.xlarge	oracle-jdk1.8.0_341_x64	1726.105	1073.354 (62.18%)	268.816 (15.57%)	46.993 (2.72%)
ecs.c7a.xlarge	oracle-jdk-11.0.16_x64	2478.146	1144.019 (46.16%)	245.967 (9.93%)	47.449 (1.91%)
ecs.c7a.xlarge	oracle-jdk-17.0.4_x64	2795.659	1082.007 (38.7%)	251.527 (9%)	52.396 (1.87%)
ecs.c7a.xlarge	oracle-jdk-18.0.2_x64	2832.787	1096.603 (38.71%)	244.491 (8.63%)	51.601 (1.82%)
ecs.c6r.xlarge	oracle-jdk1.8.0_341_aarch64	1275.118	737.23 (57.82%)	192.359 (15.09%)	36.881 (2.89%)
ecs.c6r.xlarge	oracle-jdk-11.0.16_aarch64	1665.656	794.867 (47.72%)	178.034 (10.69%)	36.896 (2.22%)
ecs.c6r.xlarge	oracle-jdk-17.0.4_aarch64	1807.769	811.996 (44.92%)	207.531 (11.48%)	38.352 (2.12%)
ecs.c6r.xlarge	oracle-jdk-18.0.2_aarch64	1738.85	829.281 (47.69%)	187.111 (10.76%)	38.478 (2.21%)
ecs.g8m.xlarge	oracle-jdk1.8.0_341_aarch64	1412.516	788.736 (55.84%)	197.661 (13.99%)	42.303 (2.99%)
ecs.g8m.xlarge	oracle-jdk-11.0.16_aarch64	2033.422	1012.32 (49.78%)	198.501 (9.76%)	39.644 (1.95%)
ecs.g8m.xlarge	oracle-jdk-17.0.4_aarch64	1237.369	1084.535 (87.65%)	208.79 (16.87%)	38.627 (3.12%)
ecs.g8m.xlarge	oracle-jdk-18.0.2_aarch64	2194.944	1077.308 (49.08%)	206.82 (9.42%)	40.76 (1.86%)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fastjson_benchmark

1. 测试环境

1.1 各个库的版本

1.2 服务器环境

ecs.c7.xlarge

ecs.c7a.xlarge

ecs.c6r.xlarge

ecs.g8m.xlarge

1.3 JDK版本

1.4 测试代码以及运行方式

2. 测试数据&结果分析

2.1 EishayParseString

2.2 EishayParseStringPretty

2.3 EishayParseTreeString

2.4 EishayParseUTF8Bytes

2.5 EishayWriteString

2.6 EishayWriteUTF8Bytes

2.7 EishayWriteBinary

2.12 EishayParseBinary

Clone this wiki locally