web应用性能测试-Tomcat 7 连接数和线程数配置

转自：http://www.jianshu.com/p/8445645b3aff

引言

这段时间折腾了哈java web应用的压力测试，部署容器是tomcat 7。期间学到了蛮多散碎的知识点，及时梳理总结，构建良好且易理解的知识架构把它们组织起来，以备忘。
对web应用开发者来说，我们很关心应用可同时处理的请求数，以及响应时间。应用本身和它运行在其中的web容器是两个很重要的影响因素。
对tomcat来说，每一个进来的请求(request)都需要一个线程，直到该请求结束。如果同时进来的请求多于当前可用的请求处理线程数，额外的线程就会被创建，直到到达配置的最大线程数(maxThreads属性值)。如果仍就同时接收到更多请求，这些来不及处理的请求就会在Connector创建的Server Socket中堆积起来，直到到达最大的配置值(acceptCount属性值)。至此，任何再来的请求将会收到connection refused错误，直到有可用的资源来处理它们。

分析、梳理、组织

这里我们关心的是tomcat能同时处理的请求数和请求响应时间，显然Connector元素的maxThreads和acceptCount属性对其有直接的影响。无论acceptCount值为多少，maxThreads直接决定了实际可同时处理的请求数。而不管maxThreads如何，acceptCount则决定了有多少请求可等待处理。然而，不管是可立即处理请求还是需要放入等待区，都需要tomcat先接受该请求(即接受client的连接请求，建立socket channel)，那么tomcat同时可建立的连接数(maxConnections属性值)也会影响可同时处理的请求数。
我们可把tomcat想象成一家医院，你来到医院大厅挂号看病，如果人家接受了，就相当于client和server建立socket连接了。接着你来到相应的科室，科室里每位医生都有一间诊室，这就相当于处理请求的线程；如果所有诊室都有病人，科室的调度护士会让你在科室小厅中耐心等待，直到他们通知你去几号诊室就诊；如果有空闲医生，你就可以立即就诊。
有的病人到医院很仓促，结果轮到他挂号或者就诊了，他还在包里翻找病例本和医保卡，如果超过了护士或医生心里可承受的等待时间，他们就会让病人到旁边找去，先服务下位。这种情形跟Connector元素的connectionTimeout属性所起的作用很相像。如果当前连接器(Connector)在接受连接后，等待了指定的时间但仍未接收到request URI line，就会抛出超时异常。

知识点收集

tomcat 7 的配置参考文档对相关属性已经描述的很详细了，这里把它们收集到一起：

protocol

Sets the protocol to handle incoming traffic. The default value is HTTP/1.1 which uses an auto-switching mechanism to select either a blocking Java based connector or an APR/native based connector. If the PATH (Windows) or LD_LIBRARY_PATH (on most unix systems) environment variables contain the Tomcat native library, the APR/native connector will be used. If the native library cannot be found, the blocking Java based connector will be used. Note that the APR/native connector has different settings for HTTPS than the Java connectors.
To use an explicit protocol rather than rely on the auto-switching mechanism described above, the following values may be used:
org.apache.coyote.http11.Http11Protocol - blocking Java connector
org.apache.coyote.http11.Http11NioProtocol - non blocking Java connector
org.apache.coyote.http11.Http11AprProtocol - the APR/native connector.
Custom implementations may also be used.
Take a look at our Connector Comparison chart. The configuration for both Java connectors is identical, for http and https.
For more information on the APR connector and APR specific SSL settings please visit the APR documentation

maxThreads

The maximum number of request processing threads to be created by this Connector, which therefore determines the maximum number of simultaneous requests that can be handled. If not specified, this attribute is set to 200. If an executor is associated with this connector, this attribute is ignored as the connector will execute tasks using the executor rather than an internal thread pool.

acceptCount

The maximum queue length for incoming connection requests when all possible request processing threads are in use. Any requests received when the queue is full will be refused. The default value is 100.

maxConnections

The maximum number of connections that the server will accept and process at any given time. When this number has been reached, the server will accept, but not process, one further connection. This additional connection be blocked until the number of connections being processed falls below maxConnections at which point the server will start accepting and processing new connections again. Note that once the limit has been reached, the operating system may still accept connections based on the acceptCount setting. The default value varies by connector type. For BIO the default is the value of maxThreads unless an Executor is used in which case the default will be the value of maxThreads from the executor. For NIO the default is 10000. For APR/native, the default is 8192.
Note that for APR/native on Windows, the configured value will be reduced to the highest multiple of 1024 that is less than or equal to maxConnections. This is done for performance reasons.
If set to a value of -1, the maxConnections feature is disabled and connections are not counted.

connectionTimeout

The number of milliseconds this Connector will wait, after accepting a connection, for the request URI line to be presented. Use a value of -1 to indicate no (i.e. infinite) timeout. The default value is 60000 (i.e. 60 seconds) but note that the standard server.xml that ships with Tomcat sets this to 20000 (i.e. 20 seconds). Unless disableUploadTimeout is set to false, this timeout will also be used when reading the request body (if any).

进一步分析

tomcat的http connector有三种：bio、nio、apr。从上面的属性描述中可以看出对于不同的connector实现，相同的属性可能会有不同的默认值和不同的处理策略，所以在调整配置前，要先弄清楚各种实现之间的不同，以及当前部署容器使用的是哪种connector。
查阅Tomcat 7 http connector 配置文档Connector Comparison部分便可获知各种connector实现间的差异。
怎样才能知道容器使用的是何种connector实现？启动tomcat后，访问Server Status Page，看到如下信息即可知道使用的是何种connector：

我的OS是windows，所以tomcat默认使用的是apr connector。在linux上，默认使用的是bio connector。与nio相比，bio性能较低。将<TOMCAT_HOME>/conf/server.xml中的如下配置片段：

<Connector port="8080" protocol="HTTP/1.1"
               connectionTimeout="20000"
               redirectPort="8443" />

修改为：

<Connector port="8080" protocol="org.apache.coyote.http11.Http11NioProtocol"
               connectionTimeout="20000"
               redirectPort="8443" />

就可将http connector切换至nio了。更多细节请参考修改Tomcat Connector运行模式，优化Tomcat运行性能