2. Who is talking?
Working in IT since 2005
● Since 2008 - Smile Ukraine CTO
● Since 2013 – OpsWay CEO
● Coaching with SmartMe.com.ua
●
3. You know this...
“Metrics show that a 0.1 second delay in page
rendering can translate into a 1% drop in
customer activity”
Amazon research papers
4. Projects are changed daily
●
Application code
●
Product database
●
Marketing tags
●
Server configuration
●
External services
●
etc
… and no QA engineers at all...
5. User checkout cycle under control
Add to cart
Select product
Success page!
Login
~30-50 sec
Billing/Shipping
options
Payment
6. Is it still running smoothly?
●
Passive checks – analyzing end-user activity
–
–
Webserver log analysis
–
●
NewRelic
Web analytics (real user monitoring)
Active checks – run test and record results
–
Gomez GPN
–
Uptrends
–
Pingdom
9. Server-side log analysis
●
Complex models:
–
2 add to cart requests timing
–
1 login step timing
–
5 checkout steps timings
●
not comparable side by side
●
not representative when user activity is low
11. Pingdom/Uptrends
●
Transaction monitoring via scripting
●
Fancy scripting language
●
●
Robust, popular but not so cheap for multiple
projects
Fine for public sites, but not for intranet
26. HAR proxy
●
Based on BrowserMob proxy server
http://bmp.lightbody.net/
HAR file
Launching WebDriver
Selenium Grid
Proxy on port 64001
Request passes
through proxy
www.mysite.com
Selenium node
(Firefox)
28. Issues we have
●
Supporting test scripts takes a lot of time
●
Support test infrastructure is tedious
●
False-positives
●
Still not enough debug information