2012 Annual State of the Union for Mobile Ecommerce Performance [Velocity EU]

Preview:

DESCRIPTION

On October 3 at Velocity EU, Strangeloop president Joshua Bixby unveiled the findings from the first study ever conducted of mobile performance over cellular networks. In July and September 2012, Strangeloop conducted an industry first: a mobile performance survey of top ecommerce sites. The "2012 State of Mobile Ecommerce Performance" documents how Strangeloop tested top Alexa-ranked retail sites on a variety of mobile devices to find answers to questions like: - How long does the median site take to load in mobile browsers? - Which sites were fastest? - Do some mobile OS/browsers/devices offer a consistently faster user experience than others? - How much faster are pages served over LTE than over 3G? - How do all of these findings compare to similar research conducted for desktop performance, published in Strangeloop’s annual Page Speed and Website Performance State of the Union reports? The report is available for download at http://www.strangeloopnetworks.com/.

Citation preview

First Annual State of the Union for Mobile Ecommerce PerformanceJoshua Bixby

www.strangeloopnetworks.comwww.webperformancetoday.com

4© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 4

5© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 5

Source: http://shuterstock.com

6© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 6

Agenda

Why speed matters

How cell networks work

Case study: What speeds up a site on mobile?

Mobile study test methodology

Results

7© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 7

Why Speed Matters

This section left intentionally blank*

*Because for the first time in five years I’m going to omit the slides that preach about speed=revenue and save all of us the repetition. If you’ve lived under a rock for the last five years and need to hear more, look at:

http://www.strangeloopnetworks.com/web-performance-optimization-hub/

8© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 8

9© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 9

Cell connectionsHow it works

10© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 10

11© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 11

http://youtu.be/LK5eHr0qdh8

12© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 12

13© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 13

14© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 14

15© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 15

http://youtu.be/ClhWIvZFS8I

16© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 16

Powering up

Source: http://www.research.att.com/articles/featured_stories/2011_03/201102_Energy_efficient?fbid=W3A01Tb8MFf

17© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 17

18© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 18

19© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 19

20© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 20

21© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 21

21© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 21

Concurrency

Device + Environment (Location, Interference etc..) + Availability + Scale

22© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 22

http://youtu.be/c1FomWJ1ZP4

23√ √√© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 23

Latency

24√ √√© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 24

Geography

25© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 25

How does data work on a mobile network?

26© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 26

3G vs LTE

27© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 27

Case StudyWhat speeds up a mobile site?

28© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 28

Methodology

29© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 29

Setup

All testing done on 3G from Vancouver.• Nexus S • Rogers Network

Traffic flows through Strangeloop service.• We apply optimizations in the path.

The phone does everything else.

30© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 30

Setup

31© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 31

Setup

32© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 32

33© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 33

34© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 34

A few points of clarification...

We’re only choosing oreilly.com for fun.

We’re using it only to show what some performance problems are. This doesn’t mean the site has these problems.

The real site is somewhere in the middle of this blue bar.

35© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 35

Summary of key metrics

36© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 36

Performance problems

Too many connections

Too many bytes

No CDN

Too many round trips

3rd party calls (marketing tags, analytics, etc)

Similar to desktop, but impact is different on mobile devices

37© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 37

What does it look like?

http://youtu.be/iPtbU1KvLjM

38© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 38

Key metrics – original site

39© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 39

Let’s Fix it!

40© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 40

What are we going to do?

Stepwise acceleration• We’ll accelerate the page in multiple steps.• We’ll add techniques at each step and evaluate the

performance impact.

Start with the easy stuff and get more advanced with each step

Try to make the page as fast as possible

41© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 41

Step 1: Low-Hanging Fruit

42© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 42

Step 1: Optimizations

Add keep-alive connections• Address the problem with too many connections• Lessen the impact of TCP connection setup

Add HTTP compression• One way to reduce bytes, and an easy one• We’ll compress the text-based content (HTML, CSS,

JS, etc)

43© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 43

Keep-alive connections

44© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 44

HTTP compression

45© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 45

Step 1: What should we expect?

Fewer total connections

Fewer bytes

Impact to performance metrics• Fewer bytes to onload• No change to roundtrips• Better document complete time• Better start render

46© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 46

Before and after

47© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 47

What we helped

Connection count• 83 connections 30 connections

Key metrics

48© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 48

Step 1: Improvements

49© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 49

Step 1: Improvements

50© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 50

What does it look like?

http://youtu.be/qMGAOqQQqvw

51© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 51

Step 2: CDN

52© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 52

CDN

53© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 53

What should we expect?

Faster average time to first byte for the objects on the page

Impact to performance metrics• No change to round trips• No change to byte count• Better start render time• Better document complete time

54© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 54

Before and after

55© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 55

What we helped

56© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 56

Step 2: Improvements

57© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 57

Step 2: Improvements

58© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 58

What does it look like?

http://youtu.be/aiViEjuVb_g

59√ √√© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 59

Why a CDN doesn’t help more: Latency

60© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 60

How does data work on a mobile network?

61© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 61

Step 3: Mobile Front End Optimization

62© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 62

Mobile FEO

Consolidation of resources• Images• CSS• JavaScript

Image compression

Minification• JavaScript• CSS

63© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 63

Mobile FEO

64© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 64

What should we expect?

Fewer round trips will mean the TTFB problem should get a lot better

Fewer bytes will mean better performance• Images (from image compression)• CSS/JS (from minification)

Impact on performance metrics• Reduction in round trips• Reduction in byte count (on top of HTTP Compression)• Better start render time• Better document complete time

65© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 65

Before and after

66© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 66

What we helped

67© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 67

Step 3: Improvements

68© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 68

Step 3: Improvements

69© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 69

What does it look like?

http://youtu.be/J_gU2Z_b-6I

70© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 70

Step 4: More Mobile FEO

71© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 71

More Mobile FEO

Deferral• Analytics• Marketing tags• Ads• Etc.

72© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 72

3rd party calls

73© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 73

3rd party calls

74© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 74

What should we expect?

Deferral of round trips to after onload

Impact to performance metrics• Fewer round trips to onload• Fewer bytes to onload• Better start render time• Better document complete time

75© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 75

Before and after

76© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 76

Before and after

77© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 77

Before and after

78© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 78

What we helped

79© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 79

Step 4: Improvements

80© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 80

Step 4: Improvements

81© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 81

What does it look like?

http://youtu.be/e0oFr0pSRn8

82© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 82

Overall improvements

83© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 83

2012 State of Mobile Ecommerce PerformanceHow fast do leading m-commerce sites load for real users over cellular networks?

84© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 84

Why is our test different?

Testing cell connecting3G testsLTE tests

Real browsers

Real phones/tablets

Cool phones/tablets

86© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 86

Report: Key areas of exploration

• Usability– Do companies have mobile-specific sites?– Who do they serve them to?– Do they let people view the full site from the mobile

site?

• Speed– How fast are web pages on a 3G connection?– How fast are web pages on an LTE connection?

• Devices– Which is faster: An Android or an iPhone?– Which is faster: An iPad or Android tablet?

85© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 85

All done by hand: Interns

87© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 87

Methodology

http://youtu.be/XapspPBxVJY

88© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 88

Usability

89© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 89

91© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 91

90© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 90

91© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 91

92© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 92

Tablets

iPad 2 versus Galaxy Tablet

Platform iOS 5.1  Android 4.04

Browser Safari  Android Browser

Processor 1 GHz Apple A5 1 GHz ARM

Display 9.5 inch 10.1-inch

Memory512 MB DDR2 (1066 Mbit/s

RAM) 1GB RAM

Cellular Radio 3G 3G

GPRS YESClass 12 (4+1/3+2/2+3/1+4

slots), 32 - 48 kbps

EDGE YES Class 12

SpeedHSDPA, 14.4 Mbps; HSUPA,

2 MbpsHSDPA, 21 Mbps; HSUPA,

5.76 Mbps

iPad 2 Galaxy Tablet

94© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 94

iPhone 4 versus Galaxy S

Platform iOS 4.1 Android 2.3.6

Browser Safari Android Browser

Processor 1GHz Apple A4 1GHz ARM

Display 3.5- inch 4-inch

Memory 512 MB eDRAM 512 MB RAM

Cellular Radio 3G 3G

GPRSClass 10 (4+1/3+2 slots), 32 -

48 kbpsClass 12 (4+1/3+2/2+3/1+4

slots), 32 - 48 kbps

EDGE Class 10, 236.8 kbps Class 12

SpeedHSDPA, 7.2 Mbps; HSUPA, 5.76 Mbps

HSDPA, 7.2 Mbps; HSUPA, 5.76 Mbps

iPhone 4 Galaxy S

96© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 96

iPhone 4 vs Galaxy S (3G)

97© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 97

iPhone 4 vs Galaxy S (3G)

98© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 98

3G compared to desktop

iPhone 5 versus Galaxy S3

Platform iOS 6 Android 4.0.4

Browser Safari Android Browser

Processor 1.3GHz Apple A6 1.4 GHz quad-core ARM

Display 4-inch 4.8-inch

Memory 1G eDRAM 1GB RAM

Cellular Radio 3G/LTE 3G, 4G

GPRS YESClass 12 (4+1/3+2/2+3/1+4 slots),

32 - 48 kbps

EDGE YES Class 12

Speed

DC-HSDPA, 42 Mbps; HSDPA, 21 Mbps; HSUPA, 5.76 Mbps, LTE, 100 Mbps;

Rev. A, up to 3.1 MbpsHSDPA, 21 Mbps; HSUPA,

5.76 Mbps

iPhone 5 Galaxy S3

100© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 100

iPhone 5 vs Galaxy S3 (LTE)

101© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 101

LTE compared to desktop

4© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 4

Source: http://fuzzyco.com/outliers/

103© 2012 Strangeloop Networks Strangeloop. Faster Websites. Automatically.© 2011 Strangeloop Networks Strangeloop. Faster Websites. Automatically. 103

Recommended