colinsurprenant

class SomeTopology < RedStorm::DSL::Topology
  configure do
    debug false

    # I reached optimal perf using one worker per node. 
    # set num_workers to N where N is the number of nodes in your cluster
    num_workers 1 # one-node cluster

    set "topology.worker.childopts", "-Djruby.compile.invokedynamic=true -Djruby.jit.threshold=0 -Djruby.jit.max=-1 -Djruby.ji.objectProxyCache=false"
  end

colinsurprenant

import org.jruby.Ruby;
import org.jruby.runtime.Helpers;
import org.jruby.runtime.builtin.IRubyObject;
import org.jruby.javasupport.JavaUtil;
import org.jruby.RubyClass;
import org.jruby.RubyModule;

public class CallJRuby {
  private static final Ruby __ruby__ = Ruby.getGlobalRuntime();

colinsurprenant

require 'benchmark'


class DSL1
  def initialize
    @i = 1
  end

  def self.register(&block)
    @block = block

colinsurprenant

# block executed using instance_exec
class DSL1
  def self.register(&block)
    @block = block
  end

  def execute(str)
    instance_exec(str, &self.class.block)
  end

colinsurprenant

EC2/Ubuntu OpenVPN server config

• open UDP port 1194 using EC2 security groups

sudo apt-get install openvpn
sudo openvpn —genkey —secret /etc/openvpn/openvpn-key.txt
sudo modprobe iptable_nat
echo 1 | sudo tee /proc/sys/net/ipv4/ip_forward
sudo iptables -t nat -A POSTROUTING -s 10.8.0.1/2 -o eth0 -j MASQUERADE

colinsurprenant

# -*- mode: ruby -*-
# vi: set ft=ruby :

Vagrant::Config.run do |config|
  config.vm.box = "quantal64_vb428"

  config.vm.customize ["modifyvm", :id, "--memory", 512]
  config.ssh.forward_agent = true

  config.vm.provision :chef_solo do |chef|

colinsurprenant

Quick bench of MRI Ruby 1.9.3-p392 vs 2.0.0-p0 vs JRuby 1.7.3 for computing 2000000 MD5, SHA-1 & FNV hashes. The goal of this benchmark is mainly to compare the ffi vs c-ext performance.

A MBP with a 3GHz Intel Dual Core i7 and OSX 10.8.2 was used. Rubies managed with rbenv.

• MD5 & SHA-1 are part of the Ruby stdlib.
• native FNV is in pure Ruby.
• ffi FNV is in C and for the exception of the first unwarmed JRuby test, all are using :buffer_in in place of :string & :save_errno => false
• c-ext FNV is in C and only run in MRI
• the JRuby warming process is done on every hash test by running it once and then benchmarking the second.

colinsurprenant

JRuby 1.7.3

benchmarking for 1000000 iterations
                         user     system      total        real
MD5:                 1.400000   0.040000   1.440000 (  1.007000)
SHA-1:               1.450000   0.010000   1.460000 (  1.043000)
native FNV A 32:     8.620000   0.050000   8.670000 (  8.197000)
native FNV A 64:    11.950000   0.050000  12.000000 ( 11.717000)
native FNV B 32:     3.290000   0.010000   3.300000 (  3.036000)
native FNV B 64:    11.670000   0.060000  11.730000 ( 11.286000)

colinsurprenant

package storm.starter;

import backtype.storm.Config;
import backtype.storm.StormSubmitter;
import backtype.storm.spout.SpoutOutputCollector;
import backtype.storm.task.TopologyContext;
import backtype.storm.topology.BasicOutputCollector;
import backtype.storm.topology.OutputFieldsDeclarer;
import backtype.storm.topology.TopologyBuilder;
import backtype.storm.topology.base.BaseBasicBolt;

colinsurprenant

source :rubygems

gem 'raad'
gem 'puma'
gem 'sinatra'