<div dir="ltr">ha!! gotcha!! finally I know why, the php5-xdebug was installed and enabled by default, didn't know was such a big impact of performance.<div><br></div><div>So this is going to the <a href="https://github.com/sebas5384/charm-drupal">charm's configuration</a>, so can be an option.</div>
<div class="gmail_extra"><br clear="all"><div><div dir="ltr"><div>Happy Sebas now.</div></div></div><br><div class="gmail_quote">2014-07-09 2:40 GMT-03:00 Sebastian <span dir="ltr"><<a href="mailto:sebas5384@gmail.com" target="_blank">sebas5384@gmail.com</a>></span>:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div><div><div dir="ltr"><div>Hey there!</div><div><br></div><div>After I did the Drupal charm, part of my plan was that the developers now can use the juju-local with lxc containers to deploy every project environment using bundles.</div>
<div><br></div><div>But! we notice a BIG difference of performance in the containers, so we run a PHP <a href="http://www.php-benchmark-script.com" target="_blank">benchmarking script</a>:</div><div><br></div><div><span style="font-size:large">Host (vbox):</span><font face="courier new, monospace"><b><br>
</b></font></div><div><font face="courier new, monospace"><b>vagrant@vagrant-ubuntu-trusty-64:~$</b></font><span style="font-family:'courier new',monospace"> php bench.php</span><br></div><div><div><font face="courier new, monospace">--------------------------------------</font></div>
<div><font face="courier new, monospace">test_math : 3.190 sec.</font></div><div><font face="courier new, monospace">test_stringmanipulation : 2.859 sec.</font></div><div><font face="courier new, monospace">test_loops : 1.702 sec.</font></div>
<div><font face="courier new, monospace">test_ifelse : 1.319 sec.</font></div><div><font face="courier new, monospace">--------------------------------------</font></div><div><font face="courier new, monospace">Total time: : <b><font color="#274e13">9.07 sec.</font></b></font></div>
</div><div><br></div><div><font size="4">Lxc container (into the vbox):</font></div><div><font face="courier new, monospace"><b>ubuntu@vagrant-local-machine-4:~$</b> php bench.php</font><br></div><div><div><font face="courier new, monospace">--------------------------------------</font></div>
<div><font face="courier new, monospace">test_math : 37.062 sec.</font></div><div><font face="courier new, monospace">test_stringmanipulation : 30.826 sec.</font></div><div><font face="courier new, monospace">test_loops : 5.749 sec.</font></div>
<div><font face="courier new, monospace">test_ifelse : 6.149 sec.</font></div><div><font face="courier new, monospace">--------------------------------------</font></div><div><font face="courier new, monospace">Total time: : <b><font color="#660000">79.786 sec.</font></b></font></div>
</div><div><br></div><div><br></div><div>As you can see, the containers created by Juju took 70 seconds more than the host.</div><div>The versions of the php where the same, and this was also tested in bare metal, to ensure that the virtualization of the host wasn't the issue.</div>
<div><br></div><div>We understand how this impact the implementation of Juju as a complete solution for devs and ops. Is not everyone that can buy a super computer to virtualize each charm, in his local notebook. So that's why I'm here, telling our story.</div>
<div><br></div><div>Maybe this is an issue that belongs to the LXC group, but I thought interesting to share this.</div><div><br></div><div>If someone can do the same benchmarkings and share the results, or any help to make things better, please!! it's more than welcome.</div>
<div><br></div><div>Cheers!,<br></div><div>Sebas.</div><div><br></div></div></div>
</div>
</div>
</blockquote></div><br></div></div>