advanced mobile optimization: how does it work? how do we measure success?
DESCRIPTION
In January 2012, Strangeloop president Joshua Bixby presented this session at the San Francisco/Silicon Valley Web Performance Meetup Group.TRANSCRIPT
ADVANCED MOBILE OPTIMIZATION
How do we measure success? How does it work?
http://www.youtube.com/watch?v=UbXAkbs9RX8
http://www.youtube.com/watch?v=g9_S492uq_A
http://www.youtube.com/watch?v=-wom2PDUidY
Retail customer
Top 200 Internet Retailer site US based Target geography: US and Europe $3 billion in revenue 30,000 employees
COMPANY BACKGROUND
http://www.youtube.com/watch?v=BV4M9yXAW9E
http://www.youtube.com/watch?v=4SbMKhhw6IU
HTML DELAY EXPERIMENT
Determine impact of server delays
GOAL
SOURCE: ENTER SOURCE HERE
Delay before sending HTML page Different experiments with different
delays Small % of traffic Monitor negative impact
METHODOLOGY
http://www.youtube.com/watch?v=UIlrnJsO_xA
Enterprise Application
Extranet application: login required Task based US based Target geography: US, Europe, and Asia Application launched in 2011
COMPANY BACKGROUND
SOURCE: ENTER SOURCE HERE
http://www.youtube.com/watch?v=CCXD9eF2ekQ
http://www.youtube.com/watch?v=yl50jpM9iJ8
http://www.youtube.com/watch?v=XgYt71_wtKc
http://www.youtube.com/watch?v=GrcZ-nwhT4A
© 2010 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 50
Advanced Mobile Acceleration
COMPONENTS OF A MOBILE WEB PAGE LOAD
*Diagram courtesy of Google.
MOBILE DELIVERY COMPLEXITY
Desktop Mobile
Screen size Big Small
Roundtrip time Low (20-50ms) High (100-300ms)
Bandwidth High Low
Navigation Mouse Touch
Processing speeds Fast Slow (10-20X slower)
MOBILE CACHE UNOPTIMIZED
? ? ???????????????
Page 1 (First View) Page 1 (Repeat View) Page 2
??????????????
?????????????
?????????????
? ?? ? ?
??????????????
??????????????
?????????????
?????????????
? ?
MOBILE CACHE OPTIMIZED
Page 1 (First View) Page 1 (Repeat View) Page 2 (First View)
SCRIPTABLE CACHE + USER TRACKING
3rd Party SLAs and timings
Defer inline scripts if they don’t meet the SLA.
e.g., If Optimost.com >200 ms, then defer.
Cancel scripts if they don’t meet the SLA.e.g., If Optimost.com >500 ms, then cancel.
SPDY
Dynamic Image Resizing
Dynamic image resizing based on
device type
5.5kb 3.8kb 1.5kb
Dramatic payload reduction=
Dynamic Payload Decision Making
Auto detect connection type and latency based on each session
Resources for next page(s)
WIFI 3G
New IE6 problem?