Garelic: Google Analytics as App Performance monitoring
1 like2,575 views
This document describes Garelic, an open source tool that allows using Google Analytics for app performance monitoring. It explains how Garelic uses Google Analytics features like User Timing and Custom Variables to track metrics like controller and ActiveRecord response times. Garelic implements this by monkey patching Rails and subscribing to ActiveSupport notifications to capture timing data and report it to Google Analytics. Some limitations are that it can't track redirects, AJAX, or jobs, and page caching breaks reports. Ideas for future work include error tracking and thresholding for slow responses only.
![How?
● Google Analytics Site Speed User Timing
_gaq.push([‘_trackTiming’, category, variable, time, opt_label, opt_sample]);
● Google Analytics Custom Variables
_gaq.push([`_setCustomVar’, index, name, value, opt_scope]);
● ActiveSupport::Notifications
ActiveSupport::Notifications.subscribe('process_action.action_controller') do |_, from, to, _, payload|
Garelic::Metrics.report(:action, :db_runtime, payload[:db_runtime] || 0)
Garelic::Metrics.report(:action, :view_runtime, payload[:view_runtime] || 0)
Garelic::Metrics.report(:action, :total_runtime, (to - from) * 1000)
Garelic::Metrics.action_identifier = "#{payload[:controller]}##{payload[:action]}"
end
ActiveSupport::Notifications.subscribe('sql.active_record') do |_, start, finish, _, payload|
type = payload[:name] || 'Other SQL'
Garelic::Metrics.report(:active_record, type, (finish - start) * 1000) if type != 'CACHE'
end](https://image.slidesharecdn.com/garelic-googleanalyticsasappperformancemonitoringrubyslava21-121026013430-phpapp01/85/Garelic-Google-Analytics-as-App-Performance-monitoring-12-320.jpg)
Garelic: Google Analytics as App Performance monitoring
- 1. Garelic: Google Analytics as App Performance Monitoring @jsuchal (@rubyslava #21)
- 2. Features
- 3. Controller: Average response times per action
- 4. Controller: Most time spent per action
- 7. Controller: Response times on timeline
- 8. ActiveRecord: Average/most time spent per model
- 9. Deployments: Average response times per deploy
- 10. and much more... compare, segment, slice & dice
- 11. Why? ● NewRelic is really great...but ○ "Database call response times" ■ Standard package = $24/month/server ○ "Transaction tracing" ■ Pro package = $149/month/server ○ basically unavailable for freelancers and small projects ● Proof-of-concept ○ 2 hours of hacking ○ 2 hours of waiting for Google Analytics
- 12. How? ● Google Analytics Site Speed User Timing _gaq.push([‘_trackTiming’, category, variable, time, opt_label, opt_sample]); ● Google Analytics Custom Variables _gaq.push([`_setCustomVar’, index, name, value, opt_scope]); ● ActiveSupport::Notifications ActiveSupport::Notifications.subscribe('process_action.action_controller') do |_, from, to, _, payload| Garelic::Metrics.report(:action, :db_runtime, payload[:db_runtime] || 0) Garelic::Metrics.report(:action, :view_runtime, payload[:view_runtime] || 0) Garelic::Metrics.report(:action, :total_runtime, (to - from) * 1000) Garelic::Metrics.action_identifier = "#{payload[:controller]}##{payload[:action]}" end ActiveSupport::Notifications.subscribe('sql.active_record') do |_, start, finish, _, payload| type = payload[:name] || 'Other SQL' Garelic::Metrics.report(:active_record, type, (finish - start) * 1000) if type != 'CACHE' end
- 13. How? ● Ruby Monkey Patching module ActionController class Metal < AbstractController::Base alias :dispatch_without_garelic :dispatch def dispatch(*args) Garelic::Metrics.reset! response = dispatch_without_garelic(*args) timing_data = Garelic.report_user_timing_from_metrics(Garelic::Metrics) _, _, chunks = response chunks.each do |chunk| chunk.gsub!(Garelic::Timing, timing_data) end response end end end
- 14. Limitations ● Cannot measure redirect, AJAX, async jobs ● Everything is visible in source code ○ Reporting stacktrace & raw SQL considered harmful ● Page caching breaks reports ● It's a hack & GA has its own limitations
- 15. Ideas for future development ● Tracking errors via GA events on error 500 pages ● Tracking only slow responses (thresholding) ● Share Google Analytics dashboards & custom reports ● Pull requests welcome! https://github. com/jsuchal/garelic
- 16. Related resources ● All Presentation Software is Broken http://www. igvita.com/2012/05/14/all-presentation-software-is-broken/ ● GA User Timing Method https://developers.google. com/analytics/devguides/collection/gajs/methods/gaJSApiUserTiming ● GA Custom Variables https://developers.google. com/analytics/devguides/collection/gajs/gaTrackingCustomVariables#setu p