From 9a514c9891c9fee58b99bec9c173a4996d7606bf Mon Sep 17 00:00:00 2001 From: Matthew Bergman Date: Wed, 17 Sep 2008 20:04:31 -0400 Subject: added benchmarking and profile documnetation v0.5 --- .../doc/guides/bechmarking and profiling/.DS_Store | Bin 0 -> 6148 bytes .../guides/bechmarking and profiling/Basics.html | 271 +++++++++ .../bechmarking and profiling/Examples/.DS_Store | Bin 0 -> 6148 bytes .../bechmarking and profiling/Examples/graph.html | 90 +++ .../bechmarking and profiling/Images/KGraph.png | Bin 0 -> 92342 bytes .../bechmarking and profiling/Images/KList.png | Bin 0 -> 20323 bytes .../guides/bechmarking and profiling/appendix.txt | 104 ++++ .../guides/bechmarking and profiling/basics.txt | 54 ++ .../bechmarking and profiling/definitions.txt | 24 + .../bechmarking and profiling/digging_deeper.txt | 1 + .../edge rails features.txt | 187 ++++++ .../guides/bechmarking and profiling/gameplan.txt | 27 + .../guides/bechmarking and profiling/preamble.html | 656 +++++++++++++++++++++ .../guides/bechmarking and profiling/preamble.txt | 46 ++ .../guides/bechmarking and profiling/rubyprof.txt | 180 ++++++ .../bechmarking and profiling/statistics.txt | 117 ++++ 16 files changed, 1757 insertions(+) create mode 100644 railties/doc/guides/bechmarking and profiling/.DS_Store create mode 100644 railties/doc/guides/bechmarking and profiling/Basics.html create mode 100644 railties/doc/guides/bechmarking and profiling/Examples/.DS_Store create mode 100644 railties/doc/guides/bechmarking and profiling/Examples/graph.html create mode 100644 railties/doc/guides/bechmarking and profiling/Images/KGraph.png create mode 100644 railties/doc/guides/bechmarking and profiling/Images/KList.png create mode 100644 railties/doc/guides/bechmarking and profiling/appendix.txt create mode 100644 railties/doc/guides/bechmarking and profiling/basics.txt create mode 100644 railties/doc/guides/bechmarking and profiling/definitions.txt create mode 100644 railties/doc/guides/bechmarking and profiling/digging_deeper.txt create mode 100644 railties/doc/guides/bechmarking and profiling/edge rails features.txt create mode 100644 railties/doc/guides/bechmarking and profiling/gameplan.txt create mode 100644 railties/doc/guides/bechmarking and profiling/preamble.html create mode 100644 railties/doc/guides/bechmarking and profiling/preamble.txt create mode 100644 railties/doc/guides/bechmarking and profiling/rubyprof.txt create mode 100644 railties/doc/guides/bechmarking and profiling/statistics.txt (limited to 'railties/doc/guides/bechmarking and profiling') diff --git a/railties/doc/guides/bechmarking and profiling/.DS_Store b/railties/doc/guides/bechmarking and profiling/.DS_Store new file mode 100644 index 0000000000..d933a198e6 Binary files /dev/null and b/railties/doc/guides/bechmarking and profiling/.DS_Store differ diff --git a/railties/doc/guides/bechmarking and profiling/Basics.html b/railties/doc/guides/bechmarking and profiling/Basics.html new file mode 100644 index 0000000000..5cf3939fd6 --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/Basics.html @@ -0,0 +1,271 @@ + + + + + + +Easy way to start on the road to Practical Benchmarking + + + +
+
+

So how do we start gathering this data? You already have been. Your logs are not just for error detection.

+
+
+
Processing MediaController#index (for 127.0.0.1 at 2008-07-17 21:30:21) [GET]
+  Session ID: BAh7BiIKZmxhc2hJQzonQWN0aW9uQ29udHJvbGxlcjo6Rmxhc2g6OkZsYXNo
+SGFzaHsABjoKQHVzZWR7AA==--cb57dad9c5e4704f0e1eddb3d498fef544faaf46
+  Parameters: {"action"=>"index", "controller"=>"media"}
+  Product Columns (0.003187)   SHOW FIELDS FROM `products`
+  Product Load (0.000597)   SELECT * FROM `products` WHERE (`products`.`name` = 'Escape Plane') LIMIT 1
+Rendering template within layouts/standard
+Rendering media/index
+  Track Load (0.001507)   SELECT * FROM `tracks` WHERE (`tracks`.product_id = 1) 
+  Track Columns (0.002280)   SHOW FIELDS FROM `tracks`
+Rendered layouts/_header (0.00051)
+Completed in 0.04310 (23 reqs/sec) | Rendering: 0.00819 (19%) | DB: 0.00757 (17%) | 200 OK [http://localhost/media]
+
+

SyslogLogger

+

SyslogLogger is a Logger work-alike that logs via syslog instead of to a file. You can add SyslogLogger to your Rails production environment to aggregate logs between multiple machines.

+

By default, SyslogLogger uses the program name ‘rails’, but this can be changed via the first argument to SyslogLogger.new.

+

NOTE! You can only set the SyslogLogger program name when you initialize SyslogLogger for the first time. This is a limitation of the way SyslogLogger uses syslog (and in some ways, a the way syslog(3) works). Attempts to change SyslogLogger’s program name after the first initialization will be ignored.

+

Sample usage with Rails +config/environment/production.rb +Add the following lines:

+
+
+
  require 'production_log/syslog_logger'
+  RAILS_DEFAULT_LOGGER = SyslogLogger.new
+config/environment.rb
+In 0.10.0, change this line:
+
+
+
+
  RAILS_DEFAULT_LOGGER = Logger.new("#{RAILS_ROOT}/log/#{RAILS_ENV}.log")
+to:
+
+
+
+
  RAILS_DEFAULT_LOGGER ||= Logger.new("#{RAILS_ROOT}/log/#{RAILS_ENV}.log")
+Other versions of Rails should have a similar change.
+
+

/etc/syslog.conf +Add the following lines:

+
+
+
 !rails
+ *.*                                             /var/log/production.log
+Then touch /var/log/production.log and signal syslogd with a HUP (killall -HUP syslogd, on FreeBSD).
+
+

/etc/newsyslog.conf +Add the following line:

+
+
+
  /var/log/production.log                 640  7     *    @T00  Z
+This creates a log file that is rotated every day at midnight, gzip’d, then kept for 7 days. Consult newsyslog.conf(5) for more details.
+
+

Now restart your Rails app. Your production logs should now be showing up in /var/log/production.log. If you have mulitple machines, you can log them all to a central machine with remote syslog logging for analysis. Consult your syslogd(8) manpage for further details.

+

A Hodel 3000 Compliant Logger for the Rest of Us

+

If you don't have access to your machines root system or just want something a bit easier to implement there is also a module developed by Geoffrey Grosenbach

+

link to module file

+

Just put the module in your lib directory and this to your environment.rb

+
+
+
require 'hodel_3000_compliant_logger'
+config.logger = Hodel3000CompliantLogger.new(config.log_path)
+
+

-Hodel 3000 Example

+
+
+
Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Parameters: {"action"=>"shipping", "controller"=>"checkout"}
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Columns (0.003155)   SHOW FIELDS FROM `books`
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Load (0.000881)   SELECT * FROM `books` WHERE (`books`.`id` = 1 AND (`books`.`sold` = 1)) 
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: ShippingAddress Columns (0.002683)   SHOW FIELDS FROM `shipping_addresses`
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Load (0.000362)   SELECT ounces FROM `books` WHERE (`books`.`id` = 1) 
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Rendering template within layouts/application
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Rendering checkout/shipping
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Load (0.000548)   SELECT * FROM `books` WHERE (sold = 0) LIMIT 3
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Author Columns (0.002571)   SHOW FIELDS FROM `authors`
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Author Load (0.000811)   SELECT * FROM `authors` WHERE (`authors`.`id` = 1) 
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Rendered store/_new_books (0.01358)
+Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Completed in 0.37297 (2 reqs/sec) | Rendering: 0.02971 (7%) | DB: 0.01697 (4%) | 200 OK [https://secure.jeffbooks/checkout/shipping]
+
+
+
+ + + diff --git a/railties/doc/guides/bechmarking and profiling/Examples/.DS_Store b/railties/doc/guides/bechmarking and profiling/Examples/.DS_Store new file mode 100644 index 0000000000..5008ddfcf5 Binary files /dev/null and b/railties/doc/guides/bechmarking and profiling/Examples/.DS_Store differ diff --git a/railties/doc/guides/bechmarking and profiling/Examples/graph.html b/railties/doc/guides/bechmarking and profiling/Examples/graph.html new file mode 100644 index 0000000000..f71d1f532b --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/Examples/graph.html @@ -0,0 +1,90 @@ + + + + + + + +

Profile Report

+ + + + + + + + + + + + +
Thread IDTotal Time
2147800.01
+ + + +

Thread 214780

+ + + + + + + + + + + + + + + +
%Total %Self Total Self Wait Child CallsNameLine
+ + + \ No newline at end of file diff --git a/railties/doc/guides/bechmarking and profiling/Images/KGraph.png b/railties/doc/guides/bechmarking and profiling/Images/KGraph.png new file mode 100644 index 0000000000..fecdcd0531 Binary files /dev/null and b/railties/doc/guides/bechmarking and profiling/Images/KGraph.png differ diff --git a/railties/doc/guides/bechmarking and profiling/Images/KList.png b/railties/doc/guides/bechmarking and profiling/Images/KList.png new file mode 100644 index 0000000000..57b3568832 Binary files /dev/null and b/railties/doc/guides/bechmarking and profiling/Images/KList.png differ diff --git a/railties/doc/guides/bechmarking and profiling/appendix.txt b/railties/doc/guides/bechmarking and profiling/appendix.txt new file mode 100644 index 0000000000..edda9607ed --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/appendix.txt @@ -0,0 +1,104 @@ +== Other Profiling Tools == + +There are a lot of great profiling tools out there. Some free, some not so free. This is a sort list detailing some of them. + +=== httperf === +http://www.hpl.hp.com/research/linux/httperf/[http://www.hpl.hp.com/research/linux/httperf/] + +A necessary tool in your arsenal. Very useful for load testing your website. + +#TODO write and link to a short article on how to use httperf. Anybody have a good tutorial availble. + + +=== Rails Analyzer === + +The Rails Analyzer project contains a collection of tools for Rails. It's open source and pretty speedy. It's not being actively worked on but is still contains some very useful tools. + +* The Production Log Analyzer examines Rails log files and gives back a report. It also includes action_grep which will give you all log results for a particular action. + +* The Action Profiler similar to Ruby-Prof profiler. + +* rails_stat which gives a live counter of requests per second of a running Rails app. + +* The SQL Dependency Grapher allows you to visualize the frequency of table dependencies in a Rails application. + +Their project homepage can be found at http://rails-analyzer.rubyforge.org/[http://rails-analyzer.rubyforge.org/] + +The one major caveat is that it needs your log to be in a different format from how rails sets it up specifically SyslogLogger. + + +==== SyslogLogger ==== + +SyslogLogger is a Logger work-alike that logs via syslog instead of to a file. You can add SyslogLogger to your Rails production environment to aggregate logs between multiple machines. + +More information can be found out at http://rails-analyzer.rubyforge.org/hacks/classes/SyslogLogger.html[http://rails-analyzer.rubyforge.org/hacks/classes/SyslogLogger.html] + +If you don't have access to your machines root system or just want something a bit easier to implement there is also a module developed by Geoffrey Grosenbach + +==== A Hodel 3000 Compliant Logger for the Rest of Us ==== + +Directions taken from +http://topfunky.net/svn/plugins/hodel_3000_compliant_logger/lib/hodel_3000_compliant_logger.rb[link to module file] + +Just put the module in your lib directory and add this to your environment.rb in it's config portion. + +------------------------------------------------------------ +require 'hodel_3000_compliant_logger' +config.logger = Hodel3000CompliantLogger.new(config.log_path) +------------------------------------------------------------- + +It's that simple. Your log output on restart should look like this. + +.Hodel 3000 Example +============================================================================ +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: +Parameters: {"action"=>"shipping", "controller"=>"checkout"} +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]:  +[4;36;1mBook Columns (0.003155) SHOW FIELDS FROM `books` +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]:  +[4;35;1mBook Load (0.000881) SELECT * FROM `books` WHERE (`books`.`id` = 1 AND (`books`.`sold` = 1))  +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]:  +[4;36;1mShippingAddress Columns (0.002683) SHOW FIELDS FROM `shipping_addresses` +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]:  +[4;35;1mBook Load (0.000362) SELECT ounces FROM `books` WHERE (`books`.`id` = 1)  +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: +Rendering template within layouts/application +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: +Rendering checkout/shipping +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]:  +[4;36;1mBook Load (0.000548) SELECT * FROM `books` +WHERE (sold = 0) LIMIT 3 +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]:  +[4;35;1mAuthor Columns (0.002571) SHOW FIELDS FROM `authors` +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: +Author Load (0.000811) SELECT * FROM `authors` WHERE (`authors`.`id` = 1)  +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: +Rendered store/_new_books (0.01358) +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: +Completed in 0.37297 (2 reqs/sec) | Rendering: 0.02971 (7%) | DB: 0.01697 (4%) | 200 OK [https://secure.jeffbooks/checkout/shipping] +============================================================================ + +=== Palmist === +An open source mysql query analyzer. Full featured and easy to work with. Also requires Hodel 3000 +http://www.flyingmachinestudios.com/projects/[http://www.flyingmachinestudios.com/projects/] + +=== New Relic === +http://www.newrelic.com/[http://www.newrelic.com/] + +Pretty nifty performance tools, pricey though. They do have a basic free +service both for when in development and when you put your application into production. Very simple installation and signup. + +#TODO more in-depth without being like an advertisement. + +=== FiveRuns === +http://www.fiveruns.com/[http://www.fiveruns.com/] + +#TODO give a bit more detail + +==== TuneUp ==== + +In their words "a new socially networked application profiling tool for Ruby on Rails developers. Designed for rapid application performance analysis in development, both privately or collaboratively with input from the community, FiveRuns TuneUp gives developers visibility into performance trouble spots and bottlenecks before they reach production." + +==== Manage ==== + +Like new relic a production monitoring tool. diff --git a/railties/doc/guides/bechmarking and profiling/basics.txt b/railties/doc/guides/bechmarking and profiling/basics.txt new file mode 100644 index 0000000000..24174efc7e --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/basics.txt @@ -0,0 +1,54 @@ +== On The Road to Optimization == +=== Looking at the log file in regards to optimization === + + You actually have been gathering data for benchmarking throughout your development cycle. Your logs are not just for error detection they also contain very useful information on how speedy your action is behaving. + +.Regular Log Output +============================================================================ +Processing MediaController#index (for 127.0.0.1 at 2008-07-17 21:30:21) [GET] + + Session ID: BAh7BiIKZmxhc2hJQzonQWN0aW9uQ29udHJvbGxlcjo6Rmxhc2g6OkZsYXNo +SGFzaHsABjoKQHVzZWR7AA==--cb57dad9c5e4704f0e1eddb3d498fef544faaf46 + + Parameters: {"action"=>"index", "controller"=>"media"} + + Product Columns (0.003187) SHOW FIELDS FROM `products` + Product Load (0.000597) SELECT * FROM `products` WHERE (`products`.`name` = 'Escape Plane') LIMIT 1 + +Rendering template within layouts/standard + +Rendering media/index + Track Load (0.001507) SELECT * FROM `tracks` WHERE (`tracks`.product_id = 1)  + Track Columns (0.002280) SHOW FIELDS FROM `tracks` + +Rendered layouts/_header (0.00051) + +*Completed in 0.04310 (23 reqs/sec) | Rendering: 0.00819 (19%) | DB: 0.00757 (17%) | 200 OK [http://localhost/media]* +============================================================================ + +What concerns us here is the last line of the action. + +Completed in 0.04310 (23 reqs/sec) gives us the amount of requests this specific action can handle. 0.04310 is the total amount of time the process to complete and 23 reqs/sec is an estimation from this. As we will see this number is not strictly valid since is a single instance of the process. But it does give you a general feel as to how the action is performing. + +Rendering: 0.00819 (19%) is the amount in milliseconds and the percentage of total time needed to complete the action for rendering the view + +DB: 0.00757 (17%) is the amount in milliseconds and the percentage of total time needed to complete the action for querying the database + +Pretty easy right. But wait 17+19 equals 36. 36%! where is the rest of the time going? The rest of the time is being spent processing the controller. Which is usually what takes up the bulk of the action + +=== Why the Log File on it's Own is not Helpful === + +So why can't we just use this to test our rails application. Technically that could work, but would be very stressful and slow. You don't have time to view your log after every request to see if your code is running quickly. Also a request that runs 100 reqs/sec might simply be an outlier and really usually runs at 20 reqs/sec. It's simply not enough information to do everything we need it to do but it's a start. + +But there is something else we must consider. + +=== A Simple Question, a Complicated Answer === + +Is Completed in 0.04310 (23 reqs/sec) a good time. Seems like it doesn't it. 43 ms does not outrageous time for a dynamic page load. But is this a dynamic page load. Maybe it was all cached. In which case this is very slow. Or maybe I'm running on five year old equipment and this is actually blazing fast for my G3. The truth is that we can't answer the question given the data. This is part of benchmarking. We need a baseline. Through comparative analysis of all your pages in your app, and an simple dynamic page for a control we can determine how fast your pages are actually running and if any of them need to be optimized. + +And now for something completely different a statistic lesson. + + + + + diff --git a/railties/doc/guides/bechmarking and profiling/definitions.txt b/railties/doc/guides/bechmarking and profiling/definitions.txt new file mode 100644 index 0000000000..3ba3cc41e0 --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/definitions.txt @@ -0,0 +1,24 @@ +== Terminology == + +=== What We Mean by Benchmarking and Profiling === + +Benchmarking: Is a process that enables comparison of inputs, processes or outputs between institutions (or parts of institutions) or within a single institution over time. + +Pretty blah right. If you are new to programing you probably have heard the term mostly in comparative reviews of computer and graphic card specs. If you done a bit of coding you've probably seen in mostly in terms of comparing one language to another or iterations of the same language. + +Benchmarking in terms of rails is a bit more specific. It entails comparing and contrasting various parts and pages of an application against one another. Mostly one is looking for how long a page requires to render, but memory consumption is also an area of concern. + +Profiling: When a page or process is seen to be problematic due to speed or memory consumption we profile it. Meaning we measures the behavior as the page or process runs, particularly the frequency and duration of function calls. The goal of profiling is not to find bugs, but to eliminate bottlenecks and establish a baseline for future regression testing. It must be engaged in a carefully controlled process of measurement and analysis. + +*What does that actually mean?* + +You have to have a clear goal for when you profile. It's very comparable to BDD where you are taking small steps towards a solution instead of trying to do it all in one large all encompassing step. A clearly defined set of expectations is essential for meaningful performance testing. For example, for a Web application, you need to know at least two things: +expected load in terms of concurrent users or HTTP connections and +acceptable response time. + +=== Where Does this Leave Us === + +Numbers and data. You benchmark to compare, your profile to fix. It's all about gaining data to analyze and using that information to better your application. The most important thing you should take away at the moment that this must be done in a systematic way. + +So the next logical question is how do we get this data. + diff --git a/railties/doc/guides/bechmarking and profiling/digging_deeper.txt b/railties/doc/guides/bechmarking and profiling/digging_deeper.txt new file mode 100644 index 0000000000..73a72e1264 --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/digging_deeper.txt @@ -0,0 +1 @@ +#TODO ADD A REAL EXAMPL OF PROFILING IN ACTION \ No newline at end of file diff --git a/railties/doc/guides/bechmarking and profiling/edge rails features.txt b/railties/doc/guides/bechmarking and profiling/edge rails features.txt new file mode 100644 index 0000000000..ede1c5831e --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/edge rails features.txt @@ -0,0 +1,187 @@ +== Performance Testing Built into Rails == + +As of June 20, 2008 edge rails has had a new type of Unit test geared towards profiling. Of course like most great things, getting it working takes bit of work. The test relies on statistics gathered from the Garbage Collection that isn't readily available from standard compiled ruby. There is a patch located at http://rubyforge.org/tracker/download.php/1814/7062/17676/3291/ruby186gc.patch[http://rubyforge.org/tracker/download.php/1814/7062/17676/3291/ruby186gc.patch] + +Also the test requires a new version of Ruby-Prof version of 0.6.1. It is not readily available at the moment and can most easily be found as a tarball on github. It's repository is located at git://github.com/jeremy/ruby-prof.git. + +What follows is a description of how to set up an alternative ruby install to use these features + +=== Compiling the Interpreter === + ++ +[source, bash] +---------------------------------------------------------------------------- +[User ~]$ mkdir rubygc +[User ~]$ wget ftp://ftp.ruby-lang.org/pub/ruby/1.8/ruby-1.8.6-p111.tar.gz +[User ~]$ tar -xzvf ruby-1.8.6-p111.tar.gz +[User ~]$ cd ruby-1.8.6-p111 +[User ruby-1.8.6-p111]$ curl http://rubyforge.org/tracker/download.php/1814/7062/17676/3291/ruby186gc.patch | patch -p0 + +#I like putting my alternative ruby builds in an opt directory, set the prefix to where ever you feel is most comfortable. + +[User ruby-1.8.6-p111]$ ./configure --prefix=/opt/rubygc +[User ruby-1.8.6-p111]$ sudo make && make install +---------------------------------------------------------------------------- + +Add the following lines in your ~/.profile or ~/.bash_login for convenience. + +---------------------------------------------------------------------------- +alias gcruby='/opt/rubygc/rubygc/bin/ruby' +alias gcrake='/opt/rubygc/rubygc/bin/rake' +alias gcgem='/opt/rubygc/rubygc/bin/gem' +alias gcirb=/opt/rubygc/rubygc/bin/irb' +alias gcrails='/opt/rubygc/rubygc/bin/rails' +---------------------------------------------------------------------------- + +=== Installing RubyGems === + +Next we need to install rubygems and rails so that we can use the interpreter properly. + ++ +[source, bash] +---------------------------------------------------------------------------- +[User ~]$ wget http://rubyforge.org/frs/download.php/38646/rubygems-1.2.0.tgz +[User ~]$ tar -xzvf rubygems-1.2.0.tgz +[User ~]$ cd rubygems-1.2.0 +[User rubygems-1.2.0]$ gcruby setup.rb +[User rubygems-1.2.0]$ cd ~ +[User ~]$ gcgem install rake +[User ~]$ gcgem install mysql +[User ~]$ gcgem install rails +---------------------------------------------------------------------------- + +If installing mysql gem fails ( like it did for me ), you will have to manually install it : + ++ +[source, bash] +---------------------------------------------------------------------------- +[User ~]$ cd /Users/lifo/rubygc/lib/ruby/gems/1.8/gems/mysql-2.7/ +[User mysql-2.7]$ gcruby extconf.rb --with-mysql-config +[User mysql-2.7]$ make && make install +---------------------------------------------------------------------------- + +=== Installing Jeremy Kemper's ruby-prof === + +We are in the home stretch. All we need now is ruby-proff 0.6.1 ++ +[source, bash] +---------------------------------------------------------------------------- +[User ~]$ git clone git://github.com/jeremy/ruby-prof.git +[User ~]$ cd ruby-prof/ +[User ruby-prof (master)]$ gcrake gem +[User ruby-prof (master)]$ gcgem install pkg/ruby-prof-0.6.1.gem +---------------------------------------------------------------------------- + +Finished, go get yourself a power drink! + +=== Ok so I lied, a few more things we need to do === + +You have everything we need to start profiling through rails Unit Testing. Unfortunately we are still missing a few files. I'm going to do the next step on a fresh Rails app, but it will work just as well on developmental 2.1 rails application. + +==== The Rails App ==== + +First I need to generate a rail app + ++ +[source, bash] +---------------------------------------------------------------------------- +[User ~]$ gcrails profiling_tester -d mysql +[User ~]$ cd profiling_tester +[User profiling_tester]$ script/generate scaffold item name:string +[User profiling_tester]$ gcrake db:create:all +[User profiling_tester]$ gcrake db:migrate +[User profiling_tester (master)]$ rm public/index.html +---------------------------------------------------------------------------- + +Now I'm going to init it as a git repository and add edge rails as a submodule to it. + ++ +[soure, bash] +---------------------------------------------------------------------------- +[User profiling_tester]$ git init +[User profiling_tester (master)]$ git submodule add git://github.com/rails/rails.git vendor/rails +---------------------------------------------------------------------------- + +Finally we want to change config.cache_classes to true in our environment.rb + +---------------------------------------------------------------------------- +config.cache_classes = true +---------------------------------------------------------------------------- + +If we don't cache classes, then the time Rails spends reloading and compiling our models and controllers will confound our results. Obviously we will try to make our test setup as similar as possible our production environment. + +=== Generating and Fixing the tests === + +Ok next we need to generate the test script. + ++ +[soure, bash] +---------------------------------------------------------------------------- +[User profiling_tester (master)]$ script/generate performance_test homepage +---------------------------------------------------------------------------- + +This will generate _test/performance/homepage_test.rb_ for you. However, as I have generated the project using Rails 2.1 gem, we'll need to manually generate one more file before we can go ahead. + +We need to put the following inside _test/performance/test_helper.rb + ++ +[source, ruby] +---------------------------------------------------------------------------- +require 'test_helper' +require 'performance_test_help' +---------------------------------------------------------------------------- + +Though this depends where you run your tests from and your system config. I myself run my tests from the Application root directory + +so instead of + ++ +[source, ruby] +---------------------------------------------------------------------------- +require 'test_helper' + +#I have + +require 'test/test_helper' +---------------------------------------------------------------------------- + +Also I needed to change homepage_test.rb to reflect this also + ++ +[source, ruby] +---------------------------------------------------------------------------- +require 'test/performance/test_helper.rb' +---------------------------------------------------------------------------- + +=== Testing === + +Rails has two types of performance testing : + +* profile - For finding out what's slow in something +* benchmark - For determining how fast is something + +In other words, you first _benchmark_ your application to find out if something is not fast. And then _profile_ it to find out what exactly is slowing it down. + +Now, if we look at the generated performance test ( one we generated using _script/generate performance_test_ ), it'll look something like : + +require 'performance/test_helper' + +class HomepageTest < ActionController::PerformanceTest + # Replace this with your real tests. + def test_homepage + get '/' + end +end + +The format looks very similar to that of an integration test. And guess what, that's what it is. But that doesn't stop you from testing your Model methods. You could very well write something like : + +require 'performance/test_helper' + +class UserModelTest < ActionController::PerformanceTest + # Replace this with your real tests. + def test_slow_find + User.this_takes_shlong_to_run + end +end + +Which is very useful way to profile individual processes. diff --git a/railties/doc/guides/bechmarking and profiling/gameplan.txt b/railties/doc/guides/bechmarking and profiling/gameplan.txt new file mode 100644 index 0000000000..578a70ebaa --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/gameplan.txt @@ -0,0 +1,27 @@ +== Get Yourself a Game Plan == + +You end up dealing with a large amount of data whenever you profile an application. It's crucial to use a rigorous approach to analyzing your application's performance else fail miserably in a vortex of numbers. This leads us to - + +=== The Analysis Process === + +I’m going to give an example methodology for conducting your benchmarking and profiling on an application. It is based on your typical scientific method. + +For something as complex as Benchmarking you need to take any methodology with a grain of salt but there are some basic strictures that you can depend on. + +Formulate a question you need to answer which is simple, tests the smallest measurable thing possible, and is exact. This is typically the hardest part of the experiment. From there some steps that you should follow are. + +* Develop a set of variables and processes to measure in order to answer this question! +* Profile based on the question and variables. Key problems to avoid when designing this experiment are: + - Confounding: Test one thing at a time, keep everything the same so you don't poison the data with uncontrolled processes. + - Cross Contamination: Make sure that runs from one test do not harm the other tests. + - Steady States: If you’re testing long running process. You must take the ramp up time and performance hit into your initial measurements. + - Sampling Error: Data should perform have a steady variance or range. If you get wild swings or sudden spikes, etc. then you must either account for the reason why or you have a sampling error. + - Measurement Error: Aka Human error, always go through your calculations at least twice to make sure there are no mathematical errors. . +* Do a small run of the experiment to verify the design. +* Use the small run to determine a proper sample size. +* Run the test. +* Perform the analysis on the results and determine where to go from there. + +Note: Even though we are using the typical scientific method; developing a hypothesis is not always useful in terms of profiling. The argument against using an hypothesis is that it will influence your experimental design and doesn’t really prove anything. + + diff --git a/railties/doc/guides/bechmarking and profiling/preamble.html b/railties/doc/guides/bechmarking and profiling/preamble.html new file mode 100644 index 0000000000..b52b8a7e74 --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/preamble.html @@ -0,0 +1,656 @@ + + + + + + +Benchmarking and Profiling Rails + + + +
+
+

You have successfully written a well tested ground breaking Social Marketing Network and have successfully deployed by Hand/Capistrano/Vlad to your shiny VPS. You are ready to go live after a few beta tests. But you're finding there is a problem. Everything was working wonderfully on your development setup, but now with fifty or a hundred people using your application at the same time things have slowed to a crawl. Pages are being dropped and mysql is giving people timeout errors. How are we going to fix this. It's time to Benchmark and Profile your application.

+

Benchmarking and Profiling is an important part of the development process that is talked about nearly enough for most beginning developers. Its hard enough learning a language and successfully writing an application. But without a firm understanding optimization, production ready apps are a near impossibility. No matter how well you code, or how much you know about a language there is always something that will trip up your application.

+

This article is my attempt to give the basic knowledge and methodology needed to optimize your application as painlessly as possible. We are are attempting this on two fronts. Both as a straight explanation and also through a real example of how benchmarking can speed up an application.

+

The main things that are covered are

+
    +
  • +

    +The basics of statistical analysis +

    +
    • +
  • +

    +Methodology behind benchmarking and profiling +

    +
    • +
  • +

    +Reading the log file for optimization +

    +
    • +
  • +

    +Performance Unit tests +

    +
    • +
  • +

    +Working with Ruby-Prof +

    +
    • +
  • +

    +HTTPREF #because you should know it +

    +
    • +
  • +

    +Overview of dedicated analysis options +

    +
    • +
+

There are a lot of areas we need to cover so lets start.

+
+
+

Terminology

+
+

What We Mean by Benchmarking and Profiling

+

Benchmarking: Is a process that enables comparison of inputs, processes or outputs between institutions (or parts of institutions) or within a single institution over time.

+

Pretty blah right. If you are new to programing you probably have heard the term mostly in comparative reviews of computer and graphic card specs. If you done a bit of coding you've probably seen in mostly in terms of comparing one language to another or iterations of the same language.

+

Benchmarking in terms of rails is a bit more specific. It entails comparing and contrasting various parts and pages of an application against one another. Mostly one is looking for how long a page requires to render, but memory consumption is also an area of concern.

+

Profiling: When a page or process is seen to be problematic due to speed or memory consumption we profile it. Meaning we measures the behavior as the page or process runs, particularly the frequency and duration of function calls. The goal of profiling is not to find bugs, but to eliminate bottlenecks and establish a baseline for future regression testing. It must be engaged in a carefully controlled process of measurement and analysis.

+

What does that actually mean?

+

You have to have a clear goal for when you profile. It's very comparable to BDD where you are taking small steps towards a solution instead of trying to do it all in one large all encompassing step. A clearly defined set of expectations is essential for meaningful performance testing. For example, for a Web application, you need to know at least two things: +expected load in terms of concurrent users or HTTP connections and +acceptable response time.

+

Where Does this Leave Us

+

Numbers and data. You benchmark to compare, your profile to fix. It's all about gaining data to analyze and using that information to better your application. The most important thing you should take away at the moment that this must be done in a systematic way.

+

So the next logical question is how do we get this data.

+
+

On The Road to Optimization

+
+

Looking at the log file in regards to optimization

+
+
+
You actually have been gathering data for benchmarking throughout your development cycle. Your logs are not just for error detection they also contain very useful information on how speedy your action is behaving.
+
+
+
Example: Regular Log Output
+
+

Processing MediaController#index (for 127.0.0.1 at 2008-07-17 21:30:21) [GET]

+
+
+
  Session ID: BAh7BiIKZmxhc2hJQzonQWN0aW9uQ29udHJvbGxlcjo6Rmxhc2g6OkZsYXNo
+SGFzaHsABjoKQHVzZWR7AA==--cb57dad9c5e4704f0e1eddb3d498fef544faaf46
+
+
+
+
Parameters: {"action"=>"index", "controller"=>"media"}
+
+
+
+
Product Columns (0.003187)   SHOW FIELDS FROM `products`
+Product Load (0.000597)   SELECT * FROM `products` WHERE (`products`.`name` = 'Escape Plane') LIMIT 1
+
+

Rendering template within layouts/standard

+

Rendering media/index + Track Load (0.001507) SELECT * FROM tracks WHERE (tracks.product_id = 1)  + Track Columns (0.002280) SHOW FIELDS FROM tracks

+

Rendered layouts/_header (0.00051)

+

Completed in 0.04310 (23 reqs/sec) | Rendering: 0.00819 (19%) | DB: 0.00757 (17%) | 200 OK [http://localhost/media]

+
+

What concerns us here is the last line of the action.

+

Completed in 0.04310 (23 reqs/sec) gives us the amount of requests this specific action can handle. 0.04310 is the total amount of time the process to complete and 23 reqs/sec is an estimation from this. As we will see this number is not strictly valid since is a single instance of the process. But it does give you a general feel as to how the action is performing.

+

Rendering: 0.00819 (19%) is the amount in milliseconds and the percentage of total time needed to complete the action for rendering the view

+

DB: 0.00757 (17%) is the amount in milliseconds and the percentage of total time needed to complete the action for querying the database

+

Pretty easy right. But wait 17+19 equals 36. 36%! where is the rest of the time going? The rest of the time is being spent processing the controller. Which is usually what takes up the bulk of the action

+

Why the Log File on it's Own is not Helpful

+

So why can't we just use this to test our rails application. Technically that could work, but would be very stressful and slow. You don't have time to view your log after every request to see if your code is running quickly. Also a request that runs 100 reqs/sec might simply be an outlier and really usually runs at 20 reqs/sec. It's simply not enough information to do everything we need it to do but it's a start.

+

But there is something else we must consider.

+

A Simple Question, a Complicated Answer

+

Is Completed in 0.04310 (23 reqs/sec) a good time. Seems like it doesn't it. 43 ms does not outrageous time for a dynamic page load. But is this a dynamic page load. Maybe it was all cached. In which case this is very slow. Or maybe I'm running on five year old equipment and this is actually blazing fast for my G3. The truth is that we can't answer the question given the data. This is part of benchmarking. We need a baseline. Through comparative analysis of all your pages in your app, and an simple dynamic page for a control we can determine how fast your pages are actually running and if any of them need to be optimized.

+

And now for something completely different a statistic lesson.

+
+

A Lession In Statistics

+
+

Adapted from a blog Article by Zed Shaw. His rant is funnier but will take longer to read. <br /> Programmers Need To Learn Statistics Or I Will Kill Them All

+

Why Learn Statistics

+

Statistics is a hard discipline. One can study it for years without fully grasping all the complexities. But its a necessary evil for coders of every level to at least know enough about statistics to know they know nothing about statistics. You can't optimize without it, and if you use it wrong, you'll just waste your time and the rest of your team's.

+

You must always question your metrics and try to demolish your supposed reasoning. Evidence and observation triumph over pure logic. Even the great Knuth once said: “Beware of bugs in the above code; I have only proved it correct, not tried it.”

+

Power-of-Ten Syndrome

+

If you done any benchmarking you have probably heard +“All you need to do is run that test [insert power-of-ten] times and then do an average.”

+

For newbs this whole power of ten comes about because we need enough data to minimize the results being contaminated by outliers. If you loaded a page five times with three of those times being around 75ms and twice 250ms you have no way of knowing the real average processing time for you page. But if we take a 1000 times and 950 are 75ms and 50 are 250ms we have a much clearer picture of the situation.

+

But this still begs the question of how you determine that 1000 is the correct number of iterations to improve the power of the experiment? (Power in this context basically means the chance that your experiment is right.)

+

The first thing that needs to be determined is how you are performing the samplings? 1000 iterations run in a massive sequential row? A set of 10 runs with 100 each? The statistics are different depending on which you do, but the 10 runs of 100 each would be a better approach. This lets you compare sample means and figure out if your repeated runs have any bias. More simply put, this allows you to see if you have a many or few outliers that might be poisoning your averages.

+

Another consideration is if a 1000 transactions is enough to get the process into a steady state after the ramp-up period? A common element of process control statistics is that all processes have a period in the beginning where the process isn’t stable. This “ramp-up” period is usually discarded when doing the analysis unless your run length has to include it. Most people think that 1000 is more than enough, but it totally depends on how the system functions. Many complex interacting systems can easily need 1000 iterations to get to a steady state, especially if performing 1000 transactions is very quick. Imagine a banking system that handles 10,000 transactions a second. I could take a good minute to get this system into a steady state, but your 1000 transaction test is only scratching the surface.

+

We can demonstrate this through R Code with similar means but different deviations.

+

Note: R is a system for statistical computation and graphics. It consists of a language plus a run-time environment with graphics, a debugger, access to certain system functions, and the ability to run programs stored in script files.

+
+
Example: R Code Input
+
+
+
+
a <- rnorm(100, 30, 5)
+b <- rnorm(100, 30, 20)
+
+
+

I construct two sets of 100 random samples from the normal distribution. Now, if I just take the average (mean or median) of these two sets they seem almost the same:

+
+
Example: Means of Sample
+
+
    +
  1. +

    +mean(a) + 30.05907 +

    +
    2. +
  2. +

    +mean(b) + 30.11601 +

    +
    3. +
  3. +

    +median(a) + 30.12729 +

    +
    4. +
  4. +

    +median(b) + 31.06874 +

    +
    5. +
+
+

They’re both around 30. So all good right? Not quite. If one looks at the outliers a different story emerges.

+
+
Example: Summary Output
+
+
    +
  1. +

    +summary(a) + Min. 1st Qu. Median Mean 3rd Qu. Max. + 13.33 27.00 30.13 30.06 33.43 47.23 +

    +
    2. +
  2. +

    +summary(b) + Min. 1st Qu. Median Mean 3rd Qu. Max. + -15.48 16.90 31.07 30.12 43.42 80.86 +

    +
    3. +
+
+

They aren't so similar now are they. Averages don't tell you everything. In fact in some cases they tell you almost nothing.

+

Don't Just Use Averages!

+

One cannot simply say my website “[insert power-of-ten] requests per second”. This is due to it being an Average. Without some form of range or variance error they are useless. Two averages can be the same, but hide massive differences in behavior. Without a standard deviation it’s not possible to figure out if the two might even be close. An even better approach (with normally distributed data) is to use a Student’s t-test to see if there are differences.

+

Note: A t-test is any statistical hypothesis test in which the test statistic has a Student's t distribution if the null hypothesis is true. It is applied when the population is assumed to be normally distributed but the sample sizes are small enough that the statistic on which inference is based is not normally distributed because it relies on an uncertain estimate of standard deviation rather than on a precisely known value. #TODO simply this to something I and the rest of the world will actually understand.

+

Let’s look at the standard deviation for our two samples:

+
+
Example: Standard Deviation
+
+
    +
  1. +

    +sd(a) + 5.562842 +

    +
    2. +
  2. +

    +sd(b) +[1] 19.09167 +

    +
    3. +
+
+

Stability is vastly different for these two samples If this were a web server performance run I’d say the second server (represented by b) has a major reliability problem. No, it’s not going to crash, but it’s performance response is so erratic that you’d never know how long a request would take. Even though the two servers perform the same on average, users will think the second one is slower because of how it seems to randomly perform.

+

The moral of the story is that if you give an average without standard deviations then you’re totally missing the entire point of even trying to measure something. A major goal of measurement is to develop a succinct and accurate picture of what’s going on, but if you don’t find out the standard deviation and do at least a couple graphs then you are not gaining anything from the process. There are other thing though that you must be aware of when testing your system. A big one is Confounding

+

Confounding

+

The idea of confounding is pretty simple: If you want to measure something, then don’t measure anything else.

+

An example. Imagine that someone tried to tell you that you needed to compare a bunch of flavors of ice cream for their taste, but that half of the tubs of creamy goodness were melted, and half were frozen. Do you think having to slop down a gallon of Heath Crunch flavored warm milk would skew your quality measurement? Of course it would. The temperature of the ice cream is confounding your comparison of taste quality. In order to fix the problem you need to remove this confounding element by having all the ice cream at a constant temperature.

+

#TODO add more information in how to avoid confounding.

+
    +
  • +

    +Your testing system and your production system must be separate. You can't profile on the same system because you are using resources to run the test that your server should be using to serve the requests. +

    +
    • +
+

Define what you are Measuring

+

Before you can measure something you really need to lay down a very concrete definition of what you’re measuring. You should also try to measure the simplest thing you can and try to avoid confounding.

+

The most important thing to determine though is how much data you can actually send to your application through it's pipe.

+

That’s all there is to performance measurement. Sure, “how much”, “data”, and “pipe” all depend on the application, but if you need 1000 requests/second processing mojo, and you can’t get your web server to push out more than 100 requests/second, then you’ll never get your JSP+EJB+Hibernate+SOAP application anywhere near good enough. If all you can shove down your DS3 is 10k/second then you’ll never get that massive 300k flash animation to your users in time to sell them your latest Gizmodo 9000.

+

#TODO add a good metaphore

+

Books Recommendations

+

He's read a lot, I'd trust him on these.

+
    +
  • +

    +Statistics; by Freedman, Pisani, Purves, and Adhikari. Norton publishers. +

    +
    • +
  • +

    +Introductory Statistics with R; by Dalgaard. Springer publishers. +

    +
    • +
  • +

    +Statistical Computing: An Introduction to Data Analysis using S-Plus; by Crawley. Wiley publishers. +

    +
    • +
  • +

    +Statistical Process Control; by Grant, Leavenworth. McGraw-Hill publishers. +

    +
    • +
  • +

    +Statistical Methods for the Social Sciences; by Agresti, Finlay. Prentice-Hall publishers. +

    +
    • +
  • +

    +Methods of Social Research; by Baily. Free Press publishers. +

    +
    • +
  • +

    +Modern Applied Statistics with S-PLUS; by Venables, Ripley. Springer publishers. +

    +
    • +
+

Back to Business

+

Now I know this was all a bit boring, but these fundamentals a necessary for understanding what we are actually doing here. Now onto the actual code and rails processes.

+

include::edge rails features.txt[]

+
+

Understanding Performance Tests Outputs

+
+

Our First Performance Test

+

So how do we profile a request.

+

One of the things that is important to us is how long it takes to render the home page - so let's make a request to the home page. Once the request is complete, the results will be outputted in the terminal.

+

In the terminal run

+

+ +[source, bash]

+
+
+
[User profiling_tester]$ gcruby tests/performance/homepage.rb
+
+

After the tests runs for a few seconds you should see something like this.

+
+
+
HomepageTest#test_homepage (19 ms warmup)
+        process_time: 26 ms
+              memory: 298.79 KB
+             objects: 1917
+
+Finished in 2.207428 seconds.
+
+

Simple but efficient.

+
    +
  • +

    +Process Time refers to amount of time necessary to complete the action. +

    +
    • +
  • +

    +memory is the amount of information loaded into memory +

    +
    • +
  • +

    +object ??? #TODO find a good definition. Is it the amount of objects put into a ruby heap for this process? +

    +
    • +
+

In addition we also gain three types of itemized log files for each of these outputs. They can be found in your tmp directory of your application.

+

The Three types are

+
    +
  • +

    +Flat File - A simple text file with the data laid out in a grid +

    +
    • +
  • +

    +Graphical File - A html colored coded version of the simple text file with hyperlinks between the various methods. Most useful is the bolding of the main processes for each portion of the action. +

    +
    • +
  • +

    +Tree File - A file output that can be use in conjunction with KCachegrind to visualize the process +

    +
    • +
+
+ + + +
+
Note
+		KCachegrind is Linux only. For Mac this means you have to do a full KDE install to have it working in your OS. Which is over 3 gigs in size. For windows there is clone called wincachegrind but it is no longer actively being developed.
+
+

Below are examples for Flat Files and Graphical Files

+

Flat Files

+
+
Example: Flat File Output Processing Time
+
+

Thread ID: 2279160 +Total: 0.026097

+
+
+
%self     total     self     wait    child    calls  name
+ 6.41      0.06     0.04     0.00     0.02      571  Kernel#===
+ 3.17      0.00     0.00     0.00     0.00      172  Hash#[]
+ 2.42      0.00     0.00     0.00     0.00       13  MonitorMixin#mon_exit
+ 2.05      0.00     0.00     0.00     0.00       15  Array#each
+ 1.56      0.00     0.00     0.00     0.00        6  Logger#add
+ 1.55      0.00     0.00     0.00     0.00       13  MonitorMixin#mon_enter
+ 1.36      0.03     0.00     0.00     0.03        1  ActionController::Integration::Session#process
+ 1.31      0.00     0.00     0.00     0.00       13  MonitorMixin#mon_release
+ 1.15      0.00     0.00     0.00     0.00        8  MonitorMixin#synchronize-1
+ 1.09      0.00     0.00     0.00     0.00       23  Class#new
+ 1.03      0.01     0.00     0.00     0.01        5  MonitorMixin#synchronize
+ 0.89      0.00     0.00     0.00     0.00       74  Hash#default
+ 0.89      0.00     0.00     0.00     0.00        6  Hodel3000CompliantLogger#format_message
+ 0.80      0.00     0.00     0.00     0.00        9  c
+ 0.80      0.00     0.00     0.00     0.00       11  ActiveRecord::ConnectionAdapters::ConnectionHandler#retrieve_connection_pool
+ 0.79      0.01     0.00     0.00     0.01        1  ActionController::Benchmarking#perform_action_without_rescue
+ 0.18      0.00     0.00     0.00     0.00       17  <Class::Object>#allocate
+
+
+

So what do these columns tell us:

+
    +
  • +

    +%self - The percentage of time spent processing the method. This is derived from self_time/total_time +

    +
    • +
  • +

    +total - The time spent in this method and its children. +

    +
    • +
  • +

    +self - The time spent in this method. +

    +
    • +
  • +

    +wait - Time processed was queued +

    +
    • +
  • +

    +child - The time spent in this method's children. +

    +
    • +
  • +

    +calls - The number of times this method was called. +

    +
    • +
  • +

    +name - The name of the method. +

    +
    • +
+

Name can be displayed three seperate ways: + #toplevel - The root method that calls all other methods + MyObject#method - Example Hash#each, The class Hash is calling the method each + * <Object:MyObject>#test - The <> characters indicate a singleton method on a singleton class. Example <Class::Object>#allocate

+

Methods are sorted based on %self. Hence the ones taking the most time and resources will be at the top.

+

So for Array#each which is calling each on the class array. We find that it processing time is 2% of the total and was called 15 times. The rest of the information is 0.00 because the process is so fast it isn't recording times less then 100 ms.

+
+
Example: Flat File Memory Output
+
+

Thread ID: 2279160 +Total: 509.724609

+
+
+
%self     total     self     wait    child    calls  name
+ 4.62     23.57    23.57     0.00     0.00       34  String#split
+ 3.95     57.66    20.13     0.00    37.53        3  <Module::YAML>#quick_emit
+ 2.82     23.70    14.35     0.00     9.34        2  <Module::YAML>#quick_emit-1
+ 1.37     35.87     6.96     0.00    28.91        1  ActionView::Helpers::FormTagHelper#form_tag
+ 1.35      7.69     6.88     0.00     0.81        1  ActionController::HttpAuthentication::Basic::ControllerMethods#authenticate_with_http_basic
+ 1.06      6.09     5.42     0.00     0.67       90  String#gsub
+ 1.01      5.13     5.13     0.00     0.00       27  Array#-
+
+
+

Very similar to the processing time format. The main difference here is that instead of calculating time we are now concerned with the amount of KB put into memory (or is it strictly the heap)

+

So for <Module::YAML>#quick_emit which is singleton method on the class YAML it uses 57.66 KB in total, 23.57 through its own actions, 6.69 from actions it calls itself and that it was called twice.

+
+
Example: Flat File Objects
+
+

Thread ID: 2279160 +Total: 6537.000000

+
+
+
%self     total     self     wait    child    calls  name
+15.16   1096.00   991.00     0.00   105.00       66  Hash#each
+ 5.25    343.00   343.00     0.00     0.00        4  Mysql::Result#each_hash
+ 4.74   2203.00   310.00     0.00  1893.00       42  Array#each
+ 3.75   4529.00   245.00     0.00  4284.00        1  ActionView::Base::CompiledTemplates#_run_erb_47app47views47layouts47application46html46erb
+ 2.00    136.00   131.00     0.00     5.00       90  String#gsub
+ 1.73    113.00   113.00     0.00     0.00       34  String#split
+ 1.44    111.00    94.00     0.00    17.00       31  Array#each-1
+
+
+
+
+
#TODO Find correct terminology for how to describe what this is exactly profiling as in are there really 2203 array objects.
+
+

Graph Files

+

While the information gleamed from flat files is very useful we still don't know which processes each method is calling. We only know how many. This is not true for a graph file. Below is a text representation of a graph file. The actual graph file is an html entity and an example of which can be found Here

+

#TODO (Handily the graph file has links both between it many processes and to the files that actually contain them for debugging. + )

+
+
Example: Graph File
+
+

Thread ID: 21277412

+
+
+
  %total   %self     total      self    children               calls   Name
+/____________________________________________________________________________/
+100.00%   0.00%      8.77      0.00      8.77                   1     #toplevel*
+                     8.77      0.00      8.77                 1/1     Object#run_primes
+/____________________________________________________________________________/
+
+
+
+
                     8.77      0.00      8.77                 1/1     #toplevel
+100.00%   0.00%      8.77      0.00      8.77                   1     Object#run_primes*
+                     0.02      0.00      0.02                 1/1     Object#make_random_array
+                     2.09      0.00      2.09                 1/1     Object#find_largest
+                     6.66      0.00      6.66                 1/1     Object#find_primes
+/____________________________________________________________________________/
+                     0.02      0.02      0.00                 1/1     Object#make_random_array
+0.18%     0.18%      0.02      0.02      0.00                   1     Array#each_index
+                     0.00      0.00      0.00             500/500     Kernel.rand
+                     0.00      0.00      0.00             500/501     Array#[]=
+/____________________________________________________________________________/
+
+
+

As you can see the calls have been separated into slices, no longer is the order determined by process time but instead from hierarchy. Each slice profiles a primary entry, with the primary entry's parents being shown above itself and it's children found below. A primary entry can be ascertained by it having values in the %total and %self columns. Here the main entry here have been bolded for connivence.

+

So if we look at the last slice. The primary entry would be Array#each_index. It takes 0.18% of the total process time and it is only called once. It is called from Object#make_random_array which is only called once. It's children are Kernal.rand which is called by it all 500 its times that it was call in this action and Arry#[]= which was called 500 times by Array#each_index and once by some other entry.

+

Tree Files

+

It's pointless trying to represent a tree file textually so here's a few pretty pictures of it's usefulness

+
KCachegrind Graph

+Graph created by KCachegrind +

+
KCachegrind List

+List created by KCachegrind +

+

#TODO Add a bit more information to this.

+
+

Getting to the Point of all of this

+
+

Now I know all of this is a bit dry and academic. But it's a very powerful tool when you know how to leverage it properly. Which we are going to take a look at in our next section

+
+

Get Yourself a Game Plan

+
+

You end up dealing with a large amount of data whenever you profile an application. It's crucial to use a rigorous approach to analyzing your application's performance else fail miserably in a vortex of numbers. This leads us to -

+

The Analysis Process

+

I’m going to give an example methodology for conducting your benchmarking and profiling on an application. It is based on your typical scientific method.

+

For something as complex as Benchmarking you need to take any methodology with a grain of salt but there are some basic strictures that you can depend on.

+

Formulate a question you need to answer which is simple, tests the smallest measurable thing possible, and is exact. This is typically the hardest part of the experiment. From there some steps that you should follow are.

+
    +
  • +

    +Develop a set of variables and processes to measure in order to answer this question! +

    +
    • +
  • +

    +Profile based on the question and variables. Key problems to avoid when designing this experiment are: +

    +
      +
    • +

      +Confounding: Test one thing at a time, keep everything the same so you don't poison the data with uncontrolled processes. +

      +
      • +
    • +

      +Cross Contamination: Make sure that runs from one test do not harm the other tests. +

      +
      • +
    • +

      +Steady States: If you’re testing long running process. You must take the ramp up time and performance hit into your initial measurements. +

      +
      • +
    • +

      +Sampling Error: Data should perform have a steady variance or range. If you get wild swings or sudden spikes, etc. then you must either account for the reason why or you have a sampling error. +

      +
      • +
    • +

      +Measurement Error: Aka Human error, always go through your calculations at least twice to make sure there are no mathematical errors. . +

      +
      • +
    +
    • +
  • +

    +Do a small run of the experiment to verify the design. +

    +
    • +
  • +

    +Use the small run to determine a proper sample size. +

    +
    • +
  • +

    +Run the test. +

    +
    • +
  • +

    +Perform the analysis on the results and determine where to go from there. +

    +
    • +
+

Note: Even though we are using the typical scientific method; developing a hypothesis is not always useful in terms of profiling. The argument against using an hypothesis is that it will influence your experimental design and doesn’t really prove anything.

+
+

Other Profiling Tools

+
+

There are a lot of great profiling tools out there. Some free, some not so free. This is a sort list detailing some of them.

+

Rails Analyzer Tools

+

SyslogLogger

+

SyslogLogger is a Logger work-alike that logs via syslog instead of to a file. You can add SyslogLogger to your Rails production environment to aggregate logs between multiple machines.

+

By default, SyslogLogger uses the program name ‘rails’, but this can be changed via the first argument to SyslogLogger.new.

+

NOTE! You can only set the SyslogLogger program name when you initialize SyslogLogger for the first time. This is a limitation of the way SyslogLogger uses syslog (and in some ways, a the way syslog(3) works). Attempts to change SyslogLogger’s program name after the first initialization will be ignored.

+

Sample usage with Rails +config/environment/production.rb +Add the following lines:

+
+
+
  require 'production_log/syslog_logger'
+  RAILS_DEFAULT_LOGGER = SyslogLogger.new
+config/environment.rb
+In 0.10.0, change this line:
+
+
+
+
  RAILS_DEFAULT_LOGGER = Logger.new("#{RAILS_ROOT}/log/#{RAILS_ENV}.log")
+to:
+
+
+
+
  RAILS_DEFAULT_LOGGER ||= Logger.new("#{RAILS_ROOT}/log/#{RAILS_ENV}.log")
+Other versions of Rails should have a similar change.
+
+

/etc/syslog.conf +Add the following lines:

+
+
+
 !rails
+ *.*                                             /var/log/production.log
+Then touch /var/log/production.log and signal syslogd with a HUP (killall -HUP syslogd, on FreeBSD).
+
+

/etc/newsyslog.conf +Add the following line:

+
+
+
  /var/log/production.log                 640  7     *    @T00  Z
+This creates a log file that is rotated every day at midnight, gzip’d, then kept for 7 days. Consult newsyslog.conf(5) for more details.
+
+

Now restart your Rails app. Your production logs should now be showing up in /var/log/production.log. If you have mulitple machines, you can log them all to a central machine with remote syslog logging for analysis. Consult your syslogd(8) manpage for further details.

+

A Hodel 3000 Compliant Logger for the Rest of Us

+

If you don't have access to your machines root system or just want something a bit easier to implement there is also a module developed by Geoffrey Grosenbach

+

link to module file

+

Just put the module in your lib directory and this to your environment.rb

+
+
+
require 'hodel_3000_compliant_logger'
+config.logger = Hodel3000CompliantLogger.new(config.log_path)
+
+
+
Example: Hodel 3000 Example
+
+

Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Parameters: {"action"⇒"shipping", "controller"⇒"checkout"} +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Columns (0.003155) SHOW FIELDS FROM books +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Load (0.000881) SELECT FROM books WHERE (books.id = 1 AND (books.sold = 1))  +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: ShippingAddress Columns (0.002683) SHOW FIELDS FROM shipping_addresses +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Load (0.000362) SELECT ounces FROM books WHERE (books.id = 1)  +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Rendering template within layouts/application +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Rendering checkout/shipping +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Book Load (0.000548) SELECT FROM books WHERE (sold = 0) LIMIT 3 +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Author Columns (0.002571) SHOW FIELDS FROM authors +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Author Load (0.000811) SELECT * FROM authors WHERE (authors.id = 1)  +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Rendered store/_new_books (0.01358) +Jul 15 11:45:43 matthew-bergmans-macbook-pro-15 rails[16207]: Completed in 0.37297 (2 reqs/sec) | Rendering: 0.02971 (7%) | DB: 0.01697 (4%) | 200 OK [https://secure.jeffbooks/checkout/shipping]

+
+
+ + + diff --git a/railties/doc/guides/bechmarking and profiling/preamble.txt b/railties/doc/guides/bechmarking and profiling/preamble.txt new file mode 100644 index 0000000000..bba8217793 --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/preamble.txt @@ -0,0 +1,46 @@ +Benchmarking and Profiling Rails +================================ +Matthew Bergman +v0.5, September 2008 + +You have successfully written a well tested ground breaking Social Marketing Network and have successfully deployed by Hand/Capistrano/Vlad to your shiny VPS. You are ready to go live after a few beta tests. But you're finding there is a problem. Everything was working wonderfully on your development setup, but now with fifty or a hundred people using your application at the same time things have slowed to a crawl. Pages are being dropped and mysql is giving people timeout errors. How are we going to fix this. It's time to Benchmark and Profile your application. + +Benchmarking and Profiling is an important part of the development process that is talked about nearly enough for most beginning developers. Its hard enough learning a language and successfully writing an application. But without a firm understanding optimization, production ready apps are a near impossibility. No matter how well you code, or how much you know about a language there is always something that will trip up your application. + +This article is my attempt to give the basic knowledge and methodology needed to optimize your application as painlessly as possible. We are are attempting this on two fronts. Both as a straight explanation and also through a real example of how benchmarking can speed up an application. + +The main things that are covered are + +* The basics of statistical analysis +* Methodology behind benchmarking and profiling +* Reading the log file for optimization +* Performance Unit tests +* Working with Ruby-Prof +* HTTPREF #because you should know it +* Overview of dedicated analysis options + +There are a lot of areas we need to cover so lets start. + + +include::definitions.txt[] + +include::basics.txt[] + +include::statistics.txt[] + +include::edge rails features.txt[] + +include::rubyprof.txt[] + +include::digging_deeper.txt[] + +include::gameplan.txt[] + +include::appendix.txt[] + + + + + + + diff --git a/railties/doc/guides/bechmarking and profiling/rubyprof.txt b/railties/doc/guides/bechmarking and profiling/rubyprof.txt new file mode 100644 index 0000000000..edf036d13e --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/rubyprof.txt @@ -0,0 +1,180 @@ +== Understanding Performance Tests Outputs == + +=== Our First Performance Test === + +So how do we profile a request. + +One of the things that is important to us is how long it takes to render the home page - so let's make a request to the home page. Once the request is complete, the results will be outputted in the terminal. + +In the terminal run + ++ +[source, bash] +---------------------------------------------------------------------------- +[User profiling_tester]$ gcruby tests/performance/homepage.rb +---------------------------------------------------------------------------- + +After the tests runs for a few seconds you should see something like this. + +---------------------------------------------------------------------------- +HomepageTest#test_homepage (19 ms warmup) + process_time: 26 ms + memory: 298.79 KB + objects: 1917 + +Finished in 2.207428 seconds. +---------------------------------------------------------------------------- + +Simple but efficient. + +* Process Time refers to amount of time necessary to complete the action. +* memory is the amount of information loaded into memory +* object ??? #TODO find a good definition. Is it the amount of objects put into a ruby heap for this process? + +In addition we also gain three types of itemized log files for each of these outputs. They can be found in your tmp directory of your application. + +*The Three types are* + +* Flat File - A simple text file with the data laid out in a grid +* Graphical File - A html colored coded version of the simple text file with hyperlinks between the various methods. Most useful is the bolding of the main processes for each portion of the action. +* Tree File - A file output that can be use in conjunction with KCachegrind to visualize the process + +NOTE: KCachegrind is Linux only. For Mac this means you have to do a full KDE install to have it working in your OS. Which is over 3 gigs in size. For windows there is clone called wincachegrind but it is no longer actively being developed. + +Below are examples for Flat Files and Graphical Files + +=== Flat Files === + +.Flat File Output Processing Time +============================================================================ +Thread ID: 2279160 +Total: 0.026097 + + %self total self wait child calls name + 6.41 0.06 0.04 0.00 0.02 571 Kernel#=== + 3.17 0.00 0.00 0.00 0.00 172 Hash#[] + 2.42 0.00 0.00 0.00 0.00 13 MonitorMixin#mon_exit + 2.05 0.00 0.00 0.00 0.00 15 Array#each + 1.56 0.00 0.00 0.00 0.00 6 Logger#add + 1.55 0.00 0.00 0.00 0.00 13 MonitorMixin#mon_enter + 1.36 0.03 0.00 0.00 0.03 1 ActionController::Integration::Session#process + 1.31 0.00 0.00 0.00 0.00 13 MonitorMixin#mon_release + 1.15 0.00 0.00 0.00 0.00 8 MonitorMixin#synchronize-1 + 1.09 0.00 0.00 0.00 0.00 23 Class#new + 1.03 0.01 0.00 0.00 0.01 5 MonitorMixin#synchronize + 0.89 0.00 0.00 0.00 0.00 74 Hash#default + 0.89 0.00 0.00 0.00 0.00 6 Hodel3000CompliantLogger#format_message + 0.80 0.00 0.00 0.00 0.00 9 c + 0.80 0.00 0.00 0.00 0.00 11 ActiveRecord::ConnectionAdapters::ConnectionHandler#retrieve_connection_pool + 0.79 0.01 0.00 0.00 0.01 1 ActionController::Benchmarking#perform_action_without_rescue + 0.18 0.00 0.00 0.00 0.00 17 #allocate +============================================================================ + +So what do these columns tell us: + + * %self - The percentage of time spent processing the method. This is derived from self_time/total_time + * total - The time spent in this method and its children. + * self - The time spent in this method. + * wait - Time processed was queued + * child - The time spent in this method's children. + * calls - The number of times this method was called. + * name - The name of the method. + +Name can be displayed three seperate ways: + * #toplevel - The root method that calls all other methods + * MyObject#method - Example Hash#each, The class Hash is calling the method each + * #test - The <> characters indicate a singleton method on a singleton class. Example #allocate + +Methods are sorted based on %self. Hence the ones taking the most time and resources will be at the top. + +So for Array#each which is calling each on the class array. We find that it processing time is 2% of the total and was called 15 times. The rest of the information is 0.00 because the process is so fast it isn't recording times less then 100 ms. + + +.Flat File Memory Output +============================================================================ +Thread ID: 2279160 +Total: 509.724609 + + %self total self wait child calls name + 4.62 23.57 23.57 0.00 0.00 34 String#split + 3.95 57.66 20.13 0.00 37.53 3 #quick_emit + 2.82 23.70 14.35 0.00 9.34 2 #quick_emit-1 + 1.37 35.87 6.96 0.00 28.91 1 ActionView::Helpers::FormTagHelper#form_tag + 1.35 7.69 6.88 0.00 0.81 1 ActionController::HttpAuthentication::Basic::ControllerMethods#authenticate_with_http_basic + 1.06 6.09 5.42 0.00 0.67 90 String#gsub + 1.01 5.13 5.13 0.00 0.00 27 Array#- +============================================================================ + +Very similar to the processing time format. The main difference here is that instead of calculating time we are now concerned with the amount of KB put into memory *(or is it strictly the heap)* + +So for #quick_emit which is singleton method on the class YAML it uses 57.66 KB in total, 23.57 through its own actions, 6.69 from actions it calls itself and that it was called twice. + +.Flat File Objects +============================================================================ +Thread ID: 2279160 +Total: 6537.000000 + + %self total self wait child calls name + 15.16 1096.00 991.00 0.00 105.00 66 Hash#each + 5.25 343.00 343.00 0.00 0.00 4 Mysql::Result#each_hash + 4.74 2203.00 310.00 0.00 1893.00 42 Array#each + 3.75 4529.00 245.00 0.00 4284.00 1 ActionView::Base::CompiledTemplates#_run_erb_47app47views47layouts47application46html46erb + 2.00 136.00 131.00 0.00 5.00 90 String#gsub + 1.73 113.00 113.00 0.00 0.00 34 String#split + 1.44 111.00 94.00 0.00 17.00 31 Array#each-1 +============================================================================ + + #TODO Find correct terminology for how to describe what this is exactly profiling as in are there really 2203 array objects. + +=== Graph Files === + +While the information gleamed from flat files is very useful we still don't know which processes each method is calling. We only know how many. This is not true for a graph file. Below is a text representation of a graph file. The actual graph file is an html entity and an example of which can be found link:Examples/graph.html[Here] + +#TODO (Handily the graph file has links both between it many processes and to the files that actually contain them for debugging. + ) + +.Graph File +============================================================================ +Thread ID: 21277412 + + %total %self total self children calls Name +/____________________________________________________________________________/ +100.00% 0.00% 8.77 0.00 8.77 1 #toplevel* + 8.77 0.00 8.77 1/1 Object#run_primes +/____________________________________________________________________________/ + + 8.77 0.00 8.77 1/1 #toplevel +100.00% 0.00% 8.77 0.00 8.77 1 Object#run_primes* + 0.02 0.00 0.02 1/1 Object#make_random_array + 2.09 0.00 2.09 1/1 Object#find_largest + 6.66 0.00 6.66 1/1 Object#find_primes +/____________________________________________________________________________/ + 0.02 0.02 0.00 1/1 Object#make_random_array +0.18% 0.18% 0.02 0.02 0.00 1 Array#each_index + 0.00 0.00 0.00 500/500 Kernel.rand + 0.00 0.00 0.00 500/501 Array#[]= +/____________________________________________________________________________/ +============================================================================ + +As you can see the calls have been separated into slices, no longer is the order determined by process time but instead from hierarchy. Each slice profiles a primary entry, with the primary entry's parents being shown above itself and it's children found below. A primary entry can be ascertained by it having values in the %total and %self columns. Here the main entry here have been bolded for connivence. + +So if we look at the last slice. The primary entry would be Array#each_index. It takes 0.18% of the total process time and it is only called once. It is called from Object#make_random_array which is only called once. It's children are Kernal.rand which is called by it all 500 its times that it was call in this action and Arry#[]= which was called 500 times by Array#each_index and once by some other entry. + +=== Tree Files === + +It's pointless trying to represent a tree file textually so here's a few pretty pictures of it's usefulness + +.KCachegrind Graph +[caption="KCachegrind graph"] +image:Images/KGraph.png[Graph created by KCachegrind] + +.KCachegrind List +[caption="KCachegrind List"] +image:Images/KList.png[List created by KCachegrind] + +#TODO Add a bit more information to this. + +== Getting to the Point of all of this == + +Now I know all of this is a bit dry and academic. But it's a very powerful tool when you know how to leverage it properly. Which we are going to take a look at in our next section + diff --git a/railties/doc/guides/bechmarking and profiling/statistics.txt b/railties/doc/guides/bechmarking and profiling/statistics.txt new file mode 100644 index 0000000000..c3b4464d4b --- /dev/null +++ b/railties/doc/guides/bechmarking and profiling/statistics.txt @@ -0,0 +1,117 @@ +== A Lession In Statistics == + +Adapted from a blog Article by Zed Shaw. His rant is funnier but will take longer to read.
http://www.zedshaw.com/rants/programmer_stats.html[Programmers Need To Learn Statistics Or I Will Kill Them All] + +=== Why Learn Statistics === + +Statistics is a hard discipline. One can study it for years without fully grasping all the complexities. But its a necessary evil for coders of every level to at least know enough about statistics to know they know nothing about statistics. You can't optimize without it, and if you use it wrong, you'll just waste your time and the rest of your team's. + +You must always question your metrics and try to demolish your supposed reasoning. Evidence and observation triumph over pure logic. Even the great Knuth once said: “Beware of bugs in the above code; I have only proved it correct, not tried it.” + +=== Power-of-Ten Syndrome === + +If you done any benchmarking you have probably heard +“All you need to do is run that test [insert power-of-ten] times and then do an average.” + +For newbs this whole power of ten comes about because we need enough data to minimize the results being contaminated by outliers. If you loaded a page five times with three of those times being around 75ms and twice 250ms you have no way of knowing the real average processing time for you page. But if we take a 1000 times and 950 are 75ms and 50 are 250ms we have a much clearer picture of the situation. + +But this still begs the question of how you determine that 1000 is the correct number of iterations to improve the power of the experiment? (Power in this context basically means the chance that your experiment is right.) + +The first thing that needs to be determined is how you are performing the samplings? 1000 iterations run in a massive sequential row? A set of 10 runs with 100 each? The statistics are different depending on which you do, but the 10 runs of 100 each would be a better approach. This lets you compare sample means and figure out if your repeated runs have any bias. More simply put, this allows you to see if you have a many or few outliers that might be poisoning your averages. + +Another consideration is if a 1000 transactions is enough to get the process into a steady state after the ramp-up period? A common element of process control statistics is that all processes have a period in the beginning where the process isn’t stable. This “ramp-up” period is usually discarded when doing the analysis unless your run length has to include it. Most people think that 1000 is more than enough, but it totally depends on how the system functions. Many complex interacting systems can easily need 1000 iterations to get to a steady state, especially if performing 1000 transactions is very quick. Imagine a banking system that handles 10,000 transactions a second. I could take a good minute to get this system into a steady state, but your 1000 transaction test is only scratching the surface. + +We can demonstrate this through R Code with similar means but different deviations. + +Note: R is a system for statistical computation and graphics. It consists of a language plus a run-time environment with graphics, a debugger, access to certain system functions, and the ability to run programs stored in script files. + +.R Code Input +============================================================================ + a <- rnorm(100, 30, 5) + b <- rnorm(100, 30, 20) +============================================================================ + +I construct two sets of 100 random samples from the normal distribution. Now, if I just take the average (mean or median) of these two sets they seem almost the same: + +.Means of Sample +============================================================================ +> mean(a) + 30.05907 +> mean(b) + 30.11601 +> median(a) + 30.12729 +> median(b) + 31.06874 +============================================================================ + +They’re both around 30. So all good right? Not quite. If one looks at the outliers a different story emerges. + +.Summary Output +============================================================================ +> summary(a) + Min. 1st Qu. Median Mean 3rd Qu. Max. + 13.33 27.00 30.13 30.06 33.43 47.23 +> summary(b) + Min. 1st Qu. Median Mean 3rd Qu. Max. + -15.48 16.90 31.07 30.12 43.42 80.86 +============================================================================ + +They aren't so similar now are they. Averages don't tell you everything. In fact in some cases they tell you almost nothing. + +=== Don't Just Use Averages! === + +One cannot simply say my website “[insert power-of-ten] requests per second”. This is due to it being an Average. Without some form of range or variance error they are useless. Two averages can be the same, but hide massive differences in behavior. Without a standard deviation it’s not possible to figure out if the two might even be close. An even better approach (with normally distributed data) is to use a Student’s t-test to see if there are differences. + +Note: A t-test is any statistical hypothesis test in which the test statistic has a Student's t distribution if the null hypothesis is true. It is applied when the population is assumed to be normally distributed but the sample sizes are small enough that the statistic on which inference is based is not normally distributed because it relies on an uncertain estimate of standard deviation rather than on a precisely known value. #TODO simply this to something I and the rest of the world will actually understand. + +Let’s look at the standard deviation for our two samples: + +.Standard Deviation +============================================================================ +> sd(a) + 5.562842 +> sd(b) +[1] 19.09167 +============================================================================ + +Stability is vastly different for these two samples If this were a web server performance run I’d say the second server (represented by b) has a major reliability problem. No, it’s not going to crash, but it’s performance response is so erratic that you’d never know how long a request would take. Even though the two servers perform the same on average, users will think the second one is slower because of how it seems to randomly perform. + +The moral of the story is that if you give an average without standard deviations then you’re totally missing the entire point of even trying to measure something. A major goal of measurement is to develop a succinct and accurate picture of what’s going on, but if you don’t find out the standard deviation and do at least a couple graphs then you are not gaining anything from the process. There are other thing though that you must be aware of when testing your system. A big one is Confounding + +=== Confounding === + +The idea of confounding is pretty simple: If you want to measure something, then don’t measure anything else. + +An example. Imagine that someone tried to tell you that you needed to compare a bunch of flavors of ice cream for their taste, but that half of the tubs of creamy goodness were melted, and half were frozen. Do you think having to slop down a gallon of Heath Crunch flavored warm milk would skew your quality measurement? Of course it would. The temperature of the ice cream is confounding your comparison of taste quality. In order to fix the problem you need to remove this confounding element by having all the ice cream at a constant temperature. + +#TODO add more information in how to avoid confounding. + +* Your testing system and your production system must be separate. You can't profile on the same system because you are using resources to run the test that your server should be using to serve the requests. + +=== Define what you are Measuring === + +Before you can measure something you really need to lay down a very concrete definition of what you’re measuring. You should also try to measure the simplest thing you can and try to avoid confounding. + +The most important thing to determine though is how much data you can actually send to your application through it's pipe. + +That’s all there is to performance measurement. Sure, “how much”, “data”, and “pipe” all depend on the application, but if you need 1000 requests/second processing mojo, and you can’t get your web server to push out more than 100 requests/second, then you’ll never get your JSP+EJB+Hibernate+SOAP application anywhere near good enough. If all you can shove down your DS3 is 10k/second then you’ll never get that massive 300k flash animation to your users in time to sell them your latest Gizmodo 9000. + +#TODO add a good metaphore + +=== Books Recommendations === +He's read a lot, I'd trust him on these. + +* Statistics; by Freedman, Pisani, Purves, and Adhikari. Norton publishers. +* Introductory Statistics with R; by Dalgaard. Springer publishers. +* Statistical Computing: An Introduction to Data Analysis using S-Plus; by Crawley. Wiley publishers. +* Statistical Process Control; by Grant, Leavenworth. McGraw-Hill publishers. +* Statistical Methods for the Social Sciences; by Agresti, Finlay. Prentice-Hall publishers. +* Methods of Social Research; by Baily. Free Press publishers. +* Modern Applied Statistics with S-PLUS; by Venables, Ripley. Springer publishers. + +=== Back to Business === + +Now I know this was all a bit boring, but these fundamentals a necessary for understanding what we are actually doing here. Now onto the actual code and rails processes. + + -- cgit v1.2.3