Benchmarking and performance tuning of Apache

By OK -

Benchmarking and performance tuning of Apache

In this recipe, we will learn some performance tuning configurations that may help to squeeze out the last bit of performance from the available hardware. Before diving into performance tuning, we need to evaluate our servers and set a benchmark which can be used to measure improvements after any changes. We will be using a well known HTTP benchmarking tool, Apache Bench (ab). Various other benchmarking tools are available and each one has its own feature set. You can choose the one that best suits your needs.

Getting ready

You will need two systems: one with the web server software installed and another to run Apache Bench. You will need root access or access to an account with similar privileges.
You will also need to modify a few network parameters to handle a large network load. You will also need to set a higher open files limit, in limits.conf, on both systems. Check the Tuning TCP Stack recipe in Chapter 2, Networking.

How to do it…

  1. Install the Apache Bench tool. This is available with the package apache2-utils:
    $ sudo apt-get install apache2-utils
  2. If you need to, you can check all the available options of the ab tool as follows:
    $ ab -h
  3. Now we are ready to generate network load. Execute the following command to start ab:
    $ ab -n 10000 -c 200 -t 2 -k "http://192.168.56.103/index.php"
    It will take some time to complete the command depending on the parameters. You should see similar results to the following (partial) output:
    How to do it…
Additionally, you may want to benchmark your server for CPU, memory, and IO performance. Check the Setting performance benchmarks recipe in Chapter 13, Performance Monitoring.
Now that we have a benchmark for server performance with stock installation, we can proceed with performance optimization. The following are some settings that are generally recommended for performance tuning:
  • Apache related settings:
    • Remove/disable any unused modules
    • Enable mod_gzip/mod_deflate
    • Turn HostnameLookups off
    • Use IP address in configuration files
    • Use persistence connection by enabling keepalive, then set keepalive timeout
    • Limit the uses of AllowOverride or completely disable it with AllowOverride none
    • Disable ExtendedStatus; this is useful while testing but not in production
  • Nginx related settings:
    • Set worker_processes to the count of your CPU cores or simply set it to auto
    • Set the number of worker_connections to test multiple values to find the best match for your servers
    • Set the keepalive_requests and keepalive_timeout values; these reduce the overhead of creating new connections
    • Enable idle connections with upstream servers by setting the keepalive value
    • Enable log buffering with buffer and flush parameters to access_log; this will reduce IO requests while logging
    • Reduce the log-level - you can set it to warn the user or display an error while in production
    • Set the sendfile directive to use an efficient sendfile() call from the operating system
    • Enable caching and compression
    • Make sure that you track the performance changes after each set of modifications; this way you will have exact knowledge regarding what worked and what not
    • You should also tune the TCP stack. The details of the TCP stack settings are covered in Chapter 2, Networking.

There's more…

Various other tools are available for benchmarking different features of the web server. The following are some well known tools, as well as a few latest additions:

See also

Comments

Post a Comment

Popular posts from this blog

HAproxy logging

By OK -
ubuntu@ubuntu:/etc/haproxy$ cat /etc/rsyslog.d/haproxy.conf # Create an additional socket in haproxy's chroot in order to allow logging via # /dev/log to chroot'ed HAProxy processes $ModLoad imudp $UDPServerRun 514 local0.* -/var/log/haproxy-0.log local1.* -/var/log/haproxy-1.log ### keep logs in localhost ## $AddUnixListenSocket /var/lib/haproxy/dev/log ------------------- ubuntu@ubuntu:/etc/haproxy$ cat /etc/haproxy/haproxy.cfg global chroot /var/lib/haproxy user haproxy group haproxy log 127.0.0.1   local0         log 127.0.0.1   local1 notice daemon frontend www     bind 192.168.40.128:8282    # haproxy public IP     default_backend as-backend    # backend used backend as-backend    balance leastconn    mode http  server as1 192.168.40.128:8082 check    # application srv 1 defaults log ...
Read more

teamcity Automatic Agent Start under Linux

By OK -
Automatic Agent Start under Linux To run agent automatically on the machine boot under Linux, configure daemon process with the  agent.sh start  command to start it and  agent.sh stop  command to stop it. Refer to an example procedure below: 1. Navigate to the services start/stop services scripts directory: cd /etc/init.d/ 2. Open the build agent service script: sudo vim buildAgent 3. Paste the following into the file : #!/bin/sh ### BEGIN INIT INFO # Provides: TeamCity Build Agent # Required-Start: $remote_fs $syslog # Required-Stop: $remote_fs $syslog # Default-Start: 2 3 4 5 # Default-Stop: 0 1 6 # Short-Description: Start build agent daemon at boot time # Description: Enable service provided by daemon. ### END INIT INFO #Provide the correct user name: USER="agentuser" case "$1" in start) su - $USER -c "cd BuildAgent/bin ; ./agent.sh start" ;; stop) su - $USER -c "cd BuildAgent/bin ; ./agent....
Read more

NFS mount add in fstab _netdev instead of default | firewall-cmd --list-all

By OK -
Redhat recommends us to add  _netdev  option for NFS file system in /etc/fstab. So that they can be mounted after starting the networking service. Alternative ways to refer to partitions: Label : LABEL=label Network ID Samba : //server/share NFS : server:/share SSHFS : sshfs#user@server:/share Device : /dev/sdxy (not recommended) Example #1 //192.168.1.134/share_name /mnt/share_name cifs username=server_user,password=server_password,_netdev 0 0 ^ Path to your share ^ Mountpoint ^ Username ^ Password ^ See note A Note A: The  _netdev  options makes sure the mount is only attempted AFTER a network connection has been established. Example #2 10.10.10.1:/vol1/fs1 /data nfs defaults,_netdev 0 0 ============================ [root@localhost ~]# firewall-cmd --list-all public   target: default   icmp-block-inversion: no   interfaces:   sources:   services: dhcpv6-client high-availabi...
Read more