SLIDE 1
Optimizing Mobile Performance
with Real User Monitoring
Brit Young
Mobile Product Manager New Relic
Optimizing Mobile Performance with Real User Monitoring Brit Young - - PowerPoint PPT Presentation
Optimizing Mobile Performance with Real User Monitoring Brit Young - - PowerPoint PPT Presentation
Optimizing Mobile Performance with Real User Monitoring Brit Young Mobile Product Manager New Relic What distinguishes a high performance mobile app? What distinguishes high performance for your mobile app? Source : XKCD Source : XKCD APM
SLIDE 2
SLIDE 3
SLIDE 4
What distinguishes a high performance mobile app?
SLIDE 5
SLIDE 6
What distinguishes high performance for your mobile app?
SLIDE 7
SLIDE 8 Source:XKCD
SLIDE 9 Source:XKCD
SLIDE 10
SLIDE 11
APM
Application Performance Monitoring
m
SLIDE 12
what could possibly go wrong?
SLIDE 13
★ UI Interactions ★ Network Requests ★ Third Party APIs ★ Database Requests ★ Real User Data
hardware?
- s version?
carrier? network type? geography? JSON data? UI path?
SLIDE 14
100
OS versions
4000
device models
600
network carriers
x x
240M
possible combinations
SLIDE 15
3
OS
10
devices
5
top carriers
x x
210
possible combinations
+2 (wifi) x2
- rientation
150 420
∞
x ?
user specific data & settings
SLIDE 16
Crashes
UI Responsiveness
Battery Use
70%
non-crashing performance issues SLIDE 17
average app crashes 3 - 5% of sessions what happens in the other 95 - 97%
SLIDE 18
01000010010! 01001001010001! 1100101000110010! 0101000101011101!
SLIDE 19
SLIDE 20
api.BankX.com lte api.BankXCalifornia.com Sensitive data
403
SLIDE 21
“The true price of anything you do is the amount of time you exchange for it.”
― Henry David Thoreau, Walden
SLIDE 22
Location, Location!
SLIDE 23
Average = 1- 2s
SLIDE 24
% of User Intent to Delete During Slow UI Responsiveness
100 0s 5s 30s
48% unforgiving 80% YIKES!
SLIDE 25
Test Coverage
Beta, A/B Testing
Performance Blind Spot
LAUNCH
Crash/Bug Report Troubleshooting
SLIDE 26
RUM
the only way to know what normal feels like to your users
Real User Monitoring
SLIDE 27
- Network
Requests
- Mobile Code
Execution
(UI & background work)
Back-end Services
Key Performance Categories
SLIDE 28
Network Performance
- HTTP Error Rate
- Avg ResponseTime
- Requests Per Minute
- Data Transfer Size
- Network Failure Rates
SLIDE 29
Mobile Code Execution
- View Load Time
- Database Request Time
- JSON Parsing Speed
- Image Load Time
- % CPU Usage
- Memory Usage
SLIDE 30
SLIDE 31
Monitor first for a baseline, validate acceptable range through user studies
SLIDE 32
Organize foreground and background work for seamless user experience
SLIDE 33
SLIDE 34
SLIDE 35
SLIDE 36
Actions & Decisions Metrics Desired Outcome
SLIDE 37
SLIDE 38
Happy Monitoring!
Brit Young
@brittanytarvin brittany@newrelic.com