05 tomcat mark thomas

8/7/2019 05 Tomcat Mark Thomas

1/37

Tomcat Optimisation& Performance Tuning

Mark Thomas

Senior ConsultantSpringSource


2/37

Who am I?

Apache Tomcat committer for over 5 years

,

Apache Tomcat PMC member

Member of the Apache Software Foundation

Member of the ASF security committee

Created the Tomcat security pages

Senior Software Engineer and Consultant at

SpringSource


3/37

Agenda

The optimisation / tuning process

logging

connectorscontent cache

Scaling Tomcat

Hints and tips


4/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


5/37

The process

Understand the system architecture

Set the performance target(s)

Measure current performance

Identify the current bottleneck

Fix the root cause of the bottleneck

Repeat until you meet the target


6/37

Common errors

Optimising code that doesn't need it

realistic data volumes

realistic user load

Lack of clear performance targets

uess ng w ere e o enec s

Fixing the symptom rather than the cause


7/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


8/37

Tomcat tuning

Applications typically account for >80% of request

processing time

Remember the tuning process

focus your efforts on the bottlenecks


9/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


10/37

Production logging

Default configuration is generic

catch-all logger logs to file and stdout

no overflow protection

logging is synchronised


11/37

Production logging

Remove duplicate logging (logging.properties)

.handlers = 1catalina.or .a ache. uli.FileHandler

becomes

to add rotation

java.util.logging.ConsoleHandler

.handlers = 1catalina.org.apache.juli.FileHandler

. . . . .

${catalina.base}/logs/catalina.%g.log1catalina.java.util.logging.FileHandler.limit = 200000001catalina.java.util.logging.FileHandler.count = 5


12/37

Production logging

Synchronous logging:

don't want disk IO to become the limiting factor

Asynchronous logging:

log queue uses memory

fall back to synchronised logging and/or drop some log messages


13/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


14/37

Connector tuning

Need to understand

TCP connections

HTTP transactionsHTTP Keep-Alive


15/37

Connector tuning

Additional considerations for load balancing

Connection pools


16/37

Which connector?

Java Blocking IO

JSSE based SSL

Native (APR)

Non-blocking

Java Non-blocking IO

JSSE based SSL


17/37

Which connector?

Requirement Preference order

Stability BIO APR NIO

SSL APR NIO BIO

Low concurrency BIO APR NIO

High concurrency-

BIO APR NIO

High concurrencyKeep-Alive

APR NIO BIO


18/37

Which connector?

Why would you use the NIO connector?

NIO is a pure Java solution

It is easy to switch between NIO and BIO with SSL


19/37

Connector tuning

maxThreads

maximum number of concurrent requests

for BIO, maximum number of open/activeconnections

400 is a good starting value

heavy CPU usage decreaselight CPU usage increase


20/37

Connector tuning

maxKeepAliveRequests

,

maximum number of HTTP requests per TCPconnection

set to 1 to disable keep alivedisable for BIO with very high concurrency, layer 4

,

enable for SSL, APR/NIO, layer 7 load balancer

Note BIO connector automatically disables keep alivewhen concurrent connections are high


21/37


22/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


23/37

Content cache tuning

Dynamic content is not cached

Configured using the element

cacheMaxSize10240

cacheTTL

5000

NIO/APR can use SEND_FILE


24/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


25/37

JVM tuning

Two key areas

Garbage collection

They are related

Remember to follow the tuning process


26/37

JVM tuning: memory

-Xms/-Xmx

Aim to set as low as possible

Setting too high wastes memory and can cause longgarbage collection pauses

-XX:NewSize -XX:NewRatio

Set to 25% to 33% of total Java heapSetting too high or too low leads to inefficientgarbage collection


27/37

JVM tuning: ideal garbage collection

Short lived objects never reach the Old Generation

collections

Long lived objects promoted to Old Generation

Long lived objects cleaned up by (rare) full garbagecollection


28/37

JVM tuning: garbage collection

GC pauses the application

Pause can range from milliseconds to seconds

The pause will impact your response time

How much does this matter?

- : ax ause s - : ax nor ause s

Set GC pause time goals

More frequent GC, shorter pauses


29/37

JVM tuning

There are many more options

http://blogs.sun.com/watt/resource/jvm-options-list.html


30/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


31/37

Scaling Tomcat

Load balancing

Fail over

Sharing state between Tomcat instances using clustering


32/37

Scaling Tomcat

Simplest configuration

2 * Tomcat instances

mod_proxy_http

Considerations

fail over


33/37

Scaling Tomcat

Stateless

load balancing algorithm

HTTP sessions will not work

Adding HTTP session support

Tomcat instance maintains HTTP session state

'Sticky sessions'

All requests for a session routed to same Tomcatinstance


34/37

Scaling Tomcat

Fail over

-

Asynchronous by default so usually used with stickysessions

Single line configuration for defaults

Will need additional configuration for production use


35/37

Agenda


logging

connectors

content cache

Scaling Tomcat

Hints and tips


36/37

Hints and tips

Load balancing / clustering

use load balancing and clustering in yourdevelopment environment

Redeployment can expose memory leaks

include this in your testing

Remember to follow the process


37/37

Questions?

Mark [email protected]

. .

05 tomcat mark thomas

Documents