236 mobile optimization-cdnetworks

Mobile Op*miza*on

Oct. 2013 / Sangjoon Ahn

Agenda §  Introduc9on

§  Market trend

§  Mobile Op9miza9on Technologies §  TCP op9miza9on §  FEO §  Image Op9miza9on §  Performance Test of FEO & Image Op9miza9on

§  Mobile CDN §  Considera9on of Mobile Delivery

§  Conclusion

Introduc9on -‐ Market Trend -‐ LTE Status

Mobile Trend §  According to Cisco's Visual Network Index report from Feb. 2013

§  Two-‐thirds of the world's mobile data traffic will be video by 2017. §  Global mobile data traffic will increase 13-‐fold between 2012 and 2017 §  In 2017, 4G will be 10 percent of connec9ons, but 45 percent of total traffic §  Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2012-‐201

7 : h[p://www.cisco.com/en/US/solu9ons/collateral/ns341/ns525/ns537/ns705/ns827/white_paper_c11-‐520862.html

MNO Status

Web Performance And User Expecta9on

Impact of 1 second delay

Web Accelera9on How web site page load 9me breaks down

First-‐mile

l  Database calls l  HTML page genera9on

Middle-‐mile

l  Retrieving page contents, including HTML, images, Javascript, etc., from origin server, across the internet

l  Delivering content over cable modem, Cellular network, etc., to end-‐user

l  Rendering page in the browser

Front-‐end 10 ~ 20 % 80 ~ 90%

Problem Addressed How it works

Contents Delivery Network (CDN) Network Latency Files cache in mul9ple edge caches tha

t are geographically dispersed

Dynamic Web Accelera9on (DWA)

Network Latency(especially dynamic content)

Op9mize rou9ng TCP op9miza9on Compression

Front-‐end Op9miza9on (FEO)

Reduce HTTP Requests, page size and browser bo[lenecks

Op9mize HTML code and page resources

Internet Internet

Backend

Last-‐mile

Mobile Op9miza9on QoE improvement for mobile user

Improve Latency

Improve Transfer rate

Improve Responsibility & Reduce Traffic

Improve QoE

QoE

[L4 improvement] TCP （Transport-‐level）Op*miza*on

[L5 improvement] Adopt SPDY/RE Technology HTTP (Session-‐level) op*miza*on

[L6 improvement] Adopt FEO Technology HTML (Presenta*on-‐level) Op*miza*on

[L7 improvement] Content Op*miza*on Video Pacing, Image Op*miza*on and etc.

[L1-‐L3 improvement] Adopt CDN (beZer performance than ISP N/W)

Today Topic

TCP Op9miza9on

§  TCP §  Designed to probe available bandwidth §  Does not use full bandwidth from start

TCP conges9on control & avoidance

§  Don’t know characteris9cs of cellular network §  Bufferbloat by large queue in mobile network §  Interference by middlebox deployed inside mobile network

§  Don’t consider about TCP characteris9cs §  overshoo9ng by TCP slow start

§  Limited Slow start, hystart

§  can detect conges9on by packet loss §  Delay-‐based conges9on control : vegas, westwood

§  Traffic control of heavy users §  Too much retransmit overhead on bandwidth limited network and congested

area network

TCP performance on mobile network is poor?

Performance Bo[leneck

PGW eNB

SGW NAT Mobile Device

Candidate : DRWA (Dynamic Receiver Window Adjustment)

Issues Bufferbloat by middle boxes

Limit bandwidth of heavy users

Issue of latency and available bandwidth

How to solve performance issues

DRWA (Dynamic Receiver Window Adjustment)

Traffic control at applica*on-‐level.

Limited Slow Start Tuning TCP parameter and Conges*on Control Algorithm

GGSN SGSN NB Mobile Device

Origin 3G

LTE

3G 4G

Network Latency 100 ~ 200 ms 50 ~ 100 ms

Available Bandwidth 2 ~ 7 Mbps 70 ~ 150 Mbps

Analysis of packet loss §  Download 200KB on bandwidth-‐limit of heavy user : 300Kbps

§  Rxt overhead : 32.89 %, Download 9me : 4.58 sec

Analysis of packet loss (cont’d) §  Download 2MB on bandwidth-‐limit of heavy user : 300Kbps

§  Rxt overhead : 16.5 %, Download 9me : 57.41 sec

TCP op9miza9on

§  Tuning §  Limited slow start for preven9ng overshoo9ng §  Tune Ini9al CWND

§  2.6.39+ defaults to 10 §  2.6.19+ can be configured via IP ROUTE

§  ip route change default via gateway_ipaddr dev eth0 initcwnd 10

§  Tune CA (Conges9on Avoidance) for sending less aggressively data

§  Tune backoff when detected packet loss

§  Improve rxt rate of bandwidth-‐limited user (heavy user) §  not easy to improve with current TCP §  Shape bandwidth by force.

Result of rxt rate of BW-‐limited user No Busy area : 10:00:00 ~ 15:30:00 Model 200K Num 300K Num 2M Num 50M Num

IS12S Normal TCP 157.7 10 142.1 10 117.2 10 Tuned TCP 100.3 10 103.0 10 108.5 10

ISW11SC Normal TCP 138.7 19 128.9 10 111.0 10 Tuned TCP 100.2 10 104.9 10 108.3 10

Busy area = 15:56:00 ~ 23:30:00 Model 200K Num 300K Num 2M Num 50M Num

IS12S Normal TCP 117.5 10 111.0 10 101.9 3 Tuned TCP 100.2 10 100.3 10 102.2 3

ISW11SC Normal TCP 116.7 11 120.0 10 110.6 10 Tuned TCP 100.2 11 102.2 10 106.2 10

B/W-‐limited phone

Field test with tuned TCP

FEO §  FEO technologies help to reduce the number of page resources required to down

load a given page and makes the browser process the page faster. §  mod_pagespeed(MPS) is an Open Source FEO solu9on from Google. As Google d

efined it : Apache module for rewri9ng web pages to reduce latency and bandwidth

Mod_pagespeed features §  Op9mize caching

§  Extend Cache, Local storage cache §  Outline CSS/javascript

§  Minimize Round Trip Times §  Combine and Inline CSS/Javascript §  Inline images, Sprite Images §  Fla[en CSS @imports §  Sharding domains

§  Minimize Payload sizes §  Op9mize image (PNG/JPG/GIF/WebP) : remove metadata of Image, resize image §  Minify CSS/Javascript §  Deduplicate inlined images

§  Op9mize browser rendering §  Defer Javascript §  Move CSS to Head §  Lazily load images §  Convert JPEG to Progressive

Web Request

§  DNS lookup to resolve the hostname to IP address §  New TCP connec9on requiring a full roundtrip to the server §  HTTP request requiring a full roundtrip to the server §  Server processing 9me

§  Network latency on mobile network is longer than that on wired network.

Mobile site performance challenges §  Remove network latency overhead.

§  Careful about using Domain sharding §  Don’t use redirect

§  Op9mize Cri9cal Rendering path §  Inline cri9cal CSS §  Remove blocking JavaScript §  Defer non-‐cri9cal assets

§  Reduce HTTP requests §  Image spri9ng §  Inline small image/CSS/JS

§  Reduce image size §  Op9mize Image

§  Load “above the fold”, not load en9re page. §  Lazy-‐load images

§  Watch out for extra round-‐trips §  Especially new connec9ons

Lab Experiment

§  Test with mod_pagespeed as an open source FEO solu9on from Google

§  Test scenario §  choose 5 sample sites : each web site for PC & mobile user §  Simulate mobile network environment using TC tool

naver

daum

gmarket

samsung

auone

Origin

mod_pagespeed

Latency : 0, 50, 100ms

TC

TC

w/o MPS　 naver daum gmarket samsung auone

First Second(Cache) First Second(Cache) First Second(Cache) First Second(Cache) First Second(Cache)

Request Count 75.8 0.1 112.3 2 256 22 199 1 43 8 Sent Kb 24.8 0.0 35.7 0.9 90.6 9.8 62.0 0.3 13.1 2.9 Receive Kb 393.5 0.6 553.1 0.8 5,434.1 7.8 1,079.2 18.5 231.3 10.6 HTML count/size(Kb) 7 51.0 0 0.0 14 55.3 1 0.3 12 39.4 0 0.0 3 48.7 1 18.5 2 17.8 1 8.3 JS count/size(Kb) 7 28.2 0 0.0 6 49.9 0 0.0 30 195.7 0 0.0 11 104.5 0 0.0 9 46.7 0 0.0 CSS count/size(Kb) 6 28.8 0 0.0 0 0.0 0 0.0 2 10.3 0 0.0 3 23.8 0 0.0 1 4.7 0 0.0

GIF count/size(Kb) 23.9 15.3 0 0.0

2 8.1 1 0.4

94 103.2 22 7.8

77 102.6 0 0.0

24 70.0 7 2.3

JPG count/size(Kb) 25 154.2 0 0.0 40 135.7 0 0.0 83 4,859.4 0 0.0 40 739.9 0 0.0 7 92.1 0 0.0 PNG count/size(Kb) 5.9 79.6 0.1 0.6 42 253.7 0 0.0 30 216.4 0 0.0 65 59.6 0 0.0 0 0.0 0 0.0 Loading Time (0ms delay) 1.0 0.3 1.4 0.8 2.6 1.1 1.3 0.2 0.7 0.3 Loading Time (50ms delay) 2.2 0.4 2.6 0.8 5.7 1.1 3.8 0.4 1.4 0.5 Loading Time (100ms delay) 3.5 0.3 4.6 0.7 10.1 1.3 6.9 0.6 2.3 0.7

　w MPS naver daum gmarket samsung auone


Request Count 45 9 78.6 21.4 178.9 62 41 5 23 5 Sent Kb 15.5 2.9 26.6 7.7 66.5 25.5 13.5 1.5 7.2 1.6 Receive Kb 373.7 61.3 512.2 132.8 1,546.1 88.5 702.3 144.5 219.0 31.7 HTML count/size(Kb) 7 60.9 7 61.0 13 98.2 12 94.5 9 74.0 9 74.0 3 144.2 3 144.2 2 31.5 2 31.2 JS count/size(Kb) 7 27.5 0 0.0 6 49.4 0 0.0 26 184.3 3 0.6 10 95.2 1 0.2 5 42.9 0 0.0 CSS count/size(Kb) 5 34.1 0 0.0 0 0.0 0 0.0 2 10.5 0 0.0 3 41.0 1 0.2 1 4.9 0 0.0 GIF count/size(Kb) 0 0.0 0 0.0 1.3 7.9 1.2 0.3 34 31.0 31 10.7 0 0.0 0 0.0 3 6.0 3 0.5 JPG count/size(Kb) 19 139.2 1 0.2 12 69.7 0 0.0 80.9 1,016.8 8 1.4 19 396.0 0 0.0 7 92.6 0 0.0 PNG count/size(Kb) 6 75.6 0 0.0 37.3 240.3 3.2 37.2 19 218.1 6 1.0 6 25.8 0 0.0 5 41.1 0 0.0 Loading Time (0ms delay) 1.0 0.5 1.5 0.9 2.7 1.5 1.9 0.8 0.7 0.4 Loading Time (50ms delay) 1.9 0.8 2.6 1.4 4.0 3.7 2.7 1.1 1.5 1.0 Loading Time (100ms delay) 3.0 1.3 4.5 1.8 6.8 2.4 4.0 1.8 2.5 1.6

Req No.

30.1% Sent Size

26.7% Receive Size

71.5% Loading Time

37.6% 　

Req No.

79.4% Sent Size

78.3%

Receive Size

35.0% Loading Time

42.4% 　

Req No.

40.6% Sent Size

37.3% Receive Size

5.0% Loading Time

15.0% 　

Req No.

30.0% Sent Size

25.5% Receive Size

7.4% Loading Time

3.3% 　

Req No.

46.5% Sent Size

45.1% Receive Size

5.3% Loading Time

7.3% 　

Test result of PC website

w/o MPS　 M_naver M_daum M-‐gmarket M_samsung auone


Request Count 30 1 31.9 1 68.8 3 47 46 59 0 Sent Kb 9.7 0.4 10.2 0.4 22.4 1.3 15.6 18.6 18.6 0.0 Receive Kb 209.2 0.3 146.6 0.3 565.2 1.1 340.0 8.5 195.5 0.0 HTML count/size(Kb) 2 20.8 0 0.0 4 19.1 0 0.0 1 4.0 0 0.0 2 5.1 1 0.2 1 9.1 0 0.0 JS count/size(Kb) 6 39.0 0 0.0 5 40.8 0 0.0 4 29.9 0 0.0 2 27.9 2 0.3 5 46.9 0 0.0 CSS count/size(Kb) 1 9.4 0 0.0 1 7.4 0 0.0 9 23.0 0 0.0 3 12.2 3 0.5 2 5.7 0 0.0 GIF count/size(Kb) 1 1.9 0 0.0 0 0.0 0 0.0 38.8 103.9 3 1.1 12 8.7 12 2.9 0 0.0 0 0.0 JPG count/size(Kb) 10 99.2 0 0.0 8.9 33.7 0 0.0 12 385.8 0 0.0 14 265.9 14 2.0 1 2.2 0 0.0 PNG count/size(Kb) 8 35.4 0 0.0 10 42.9 1 0.3 3 6.7 0 0.0 14 20.1 14 2.6 44 117.5 0 0.0 Loading Time (0ms delay) 1.0 0.6 0.5 0.1 0.5 0.1 0.5 0.2 0.7 0.2 Loading Time (50ms delay) 1.9 0.7 1.4 0.2 1.5 0.2 1.4 0.8 1.6 0.2 Loading Time (100ms delay) 2.8 0.8 2.4 0.3 3.0 0.3 2.3 1.4 2.7 0.2

　w MPS M_naver M_daum M-‐gmarket M_samsung auone


Request Count 28 15 28 5 36 30 18 18 40 28 Sent Kb 9.4 5.8 9.7 1.9 11.8 12.0 5.9 7.0 13.1 11.4 Receive Kb 156.4 17.2 143.0 22.5 506.4 43.1 219.4 12.8 187.1 29.8 HTML count/size(Kb) 2 14.9 2 14.7 4 22.2 3 21.8 1 37.2 1 37.2 2 11.1 2 11.0 1 25.2 1 25.2 JS count/size(Kb) 6 39.2 2 0.4 5 40.5 0 0.0 3 30.1 0 0.0 2 27.0 2 0.2 4 44.2 0 0.0 CSS count/size(Kb) 1 9.5 0 0.0 1 7.4 0 0.0 8 23.1 8 1.5 3 36.7 3 0.4 1 9.5 0 0.0 GIF count/size(Kb) 1 1.9 1 0.2 0 0.0 0 0.0 10 22.0 9 2.3 0 0.0 0 0.0 0 0.0 0 0.0 JPG count/size(Kb) 9 56.1 1 0.2 5 17.3 0 0.0 12 386.7 12 2.0 10 142.0 10 1.1 1 4.5 0 0.0 PNG count/size(Kb) 7 33.3 7 1.2 11 54.7 0 0.0 2 7.3 0 0.0 1 2.6 1 0.1 27 89.2 21 3.5 Loading Time (0ms delay) 1.1 0.7 0.6 20.0 0.5 0.2 0.3 0.2 0.7 0.3 Loading Time (50ms delay) 1.9 1.2 1.3 0.7 1.5 0.8 1.4 0.8 1.7 0.8 Loading Time (100ms delay) 2.8 1.7 2.3 1.3 2.6 1.5 2.3 1.4 2.7 1.4

Req No.

6.7% Sent Size

2.6% Receive Size

25.2% Loading Time

0% 　

Req No.

12.2% Sent Size

4.9% Receive Size

2.5% Loading Time

5.0% 　

Req No.

47.7% Sent Size

47.5% Receive Size

10.4% Loading Time

14.0% 　

Req No.

61.7% Sent Size

62.1% Receive Size

35.5% Loading Time

0% 　

Req No.

32.2% Sent Size

29.3% Receive Size

4.3% Loading Time

0% 　

Test result of Mobile website

Default tuning Aggressive tuning inline more js : 2048 -‐> 20480

inline more image : 2048 -‐> 6000 Req No.

32.2% Sent Size

29.3% Receive Size

4.3% Loading Time

0% 　

Req No.

47.5% Sent Size

46.3% Receive Size

5.4% Loading Time

28.6% 　

Mobile site

Experiment by tuning

Default tuning

Aggressive tuning fla[en_css_imports :

JpegRecompressionQuality: 100% -‐> 75%

Req No.

61.7% Sent Size

62.1% Receive Size

35.5% Loading Time

0% 　 Req No.

66.1% Sent Size

66.1% Receive Size

66.0% Loading Time

33.7% 　

Experiment by tuning (Cont’d)

Mobile site

w/o MPS　 M_samsung

First Second(Cache)

Request Count 47 46 Sent Kb 15.6 18.6 Receive Kb 340.0 8.5 HTML count/size(Kb) 2 5.1 1 0.2 JS count/size(Kb) 2 27.9 2 0.3 CSS count/size(Kb) 3 12.2 3 0.5 GIF count/size(Kb) 12 8.7 12 2.9 JPG count/size(Kb) 14 265.9 14 2.0 PNG count/size(Kb) 14 20.1 14 2.6 Loading Time (0ms delay) 0.5 0.2 Loading Time (50ms delay) 1.4 0.8 Loading Time (100ms delay) 2.3 1.4 118K

Image Op9miza9on

Background

Image is 61.8% of total size.

Mar 2013 Source : HTTP Archive §  The average web page will hit 2 MB by 201

5. §  Image is already a significant por9on of we

b page. §  61.8% of total web size is image at Mar 201

3. (826KB/1335KB) §  62.5% at Mar 2012. (630KB/1008KB) §  57.4% at Mar 2011. (445KB/775KB)

How to op9mize image §  Op9mize image formats §  Op9mize image delivery

§  U9lize Browser cache §  Use CDN

§  Op9mize image loading process §  Lazy loading

§  Op9mize image for mobile §  Responsive Images : RWD (Responsive Web Design)

Op9mize image formats §  GIF §  PNG

§  Convert from PNG 24 -‐> 8 §  PNG Op9miza9on tools : PNGCrush, Op9PNG

§  JPEG §  Control Quality §  Remove metadata §  Op9miza9on Tools : jpegtran, ImageMagick

§  WebP §  a new image format that provides lossless and lossy compression for images

on the web §  Lossless WebP 26% smaller than PNG §  Lossy WebP 25% ~ 34% smaller than JPEG §  Support browser (~26%): Chrome 9+, Android 4+, Opera 12+, Opera Mobile 1

1.1+

Image Op9miza9on Item Descrip9on

§  Recompress

§  Recompress JPEG(lossy), PNG, WEBP(lossy) images §  compression quality 0~100, -‐1(lossless) (re-‐compressing jpeg and webp images) §  Strip color profile informa9on §  Strip unnecessary meta-‐data (such as thumbnails) §  Reduc9on the color sampling of jpeg images

§  Transforma9on

§  Op9mize GIF image by conver9ng(lossless) to a PNG(except animated gifs) §  Conver9ng GIF or PNG image to a jpeg (When it hasn’t alpha channel or transparent pixels)

§  Conver9ng JPEG to WEBP §  Transforma9on larger non-‐progressive jpeg images into progressive(fade-‐in) jpegs

§  Image Resizing §  image resizing depending on screen size

Difference between FEO and Image Opt.

Request h[p://www.foo.com/index.html

Origin FEO (mod_pagespeed)

Request h[p://image.foo.com/a.gif.pagespeed.ic.4f4XMOxRCY.png

Response h[p://image.foo.com/index.html

Origin Image Opt. Request h[p://image.foo.com/a.gif

Response h[p://image.foo.com/a.gif

-‐ MIME TYPE : image/PNG

Request h[p://image.foo.com/a.gif

<html> <script src="js/jquery.js" type="text/javascript"></script> <img src=“a.gif" alt="" width="45" height="44"> </html> <html>

<script src="js/jquery.js.pagespeed.jm.LQy2C9DQIS.js" type="text/javascript"></script> <img src=“a.gif.pagespeed.ic.4f4XMOxRCY.png" alt="" width="45" height="44"> </html>

Response h[p://image.foo.com/a.gif.pagespeed.ic.4f4XMOxRCY.png

HTML Reforma*ng

Content transfo

rma*on

Image transforma*on

Image re-‐encoding from gif to png

Difference between FEO and Image Opt.(Cont’d)

Sta9c Image

JS / CSS

Traffic Reduc*on

HTML

Dynamic Image

Sta9c Image

JS / CSS

HTML

Dynamic Image

Sta9c Image

JS / CSS

HTML

Dynamic Image

Traffic Reduc*on

Traffic Reduc*on

Origin’s Web Content By FEO(mod_pagespeed) By Image Op9miza9on

Image op9miza9on

Content Minifica9on

Image op9miza9on

Image op9miza9on

Traffic Reduc*on

Experiment of Image Opt. in the field §  Experiment of image op9miza9on using mod_pagespeed on real sit

e Origin Data Image Op*miza*on by mod_pagespeed

Total Data Size(Kbyte)

Total Image Size(Kbyte)

Total Data Size(Kbyte)

Total Image Size(Kbyte)

Total Opt. ra9o

Image Opt. ra9o

Processed Image ra9o

PC Naver 1,329.9 346.6 1,262.0 263.6 5.1% 23.9% 50.6%

Daum 1,382.1 278.3 1,309.6 225.6 5.2% 18.9% 26.0%

Gmarket 3,264.8 2,989.5 3,176.9 2,923.9 2.7% 2.2% 10.2%

Samsung 1,125.6 1,003.6 453.1 341.8 59.8% 65.9% 96.6%

Auone 279.5 131.6 228.1 86.4 18.4% 34.4% 95.2%

Mobile Naver 453.9 302.8 456.8 307.4 0.0% -‐1.5% 3.1%

Daum 409.4 314.9 365.8 285.0 10.7% 9.5% 19.0%

Gmarket 869.5 802.7 497.7 439.7 42.8% 45.2% 99.9%

Samsung 368.1 314.7 117.5 64.9 68.1% 79.4% 99.7%

Auone 213.4 132.3 172.9 98.4 19.0% 25.7% 90.5%

§  Processed Image Ra9o = (Processed Image size / Total origin image size) §  We can know how much image data can be op9mized.

We can know how much image is op9mized

If op9mize dynamic image, we can reduce more image traffic.

Experiment of Image Opt. in the field §  Experiment of image op9miza9on in the field

Origin Data mod_pagespeed Image_op*mizer

Total Data Size(KB)

Total Image Size(KB)

Total Data Size(KB)


Total Opt. ra9o

Image Opt. ra9o


Total Data Size(KB)


Total Opt. ra9o

Image Opt. ra9o


PC Naver 1366.9 436.8 1251.9 367.2 8% 16% 73% 1216.4 313.1 11% 28% 98% Daum 1110.6 278.5 1116.8 240.4 -‐1% 14% 88% 1064.5 233.1 4% 16% 86% Gmarket 8244.7 7888.8 7453.9 7124.6 10% 10% 25% 1428.4 1078.9 83% 86% 101% Samsung 6177.1 5396.7 1408.3 627.9 77% 88% 70% 1384.5 603.5 78% 89% 89% Auone 343.5 123.8 302 85.8 12% 31% 90% 305.1 84.7 11% 32% 100% Mobage 606.9 337.2 574.4 314.3 5% 7% 35% 469 211.0 23% 37% 95% Gree 417.7 326.1 420.7 319.1 -‐1% 2% 0% 264.5 173.0 37% 47% 100%

Mobile

Naver 340.1 209.4 321.3 200.1 6% 4% 42% 330.9 200.1 3% 4% 38% Daum 298.9 161.2 283.6 160.9 5% 0% 0% 287.1 149.3 4% 7% 72% Gmarket 2519.3 2368.8 1208.1 1055.3 52% 55% 85% 1183.5 1014.8 53% 57% 87% Samsung 438.5 262.9 229.9 54.3 48% 79% 88% 227.7 52.0 48% 80% 86% Auone 585.9 358.4 531.5 322.2 9% 10% 21% 526.8 299.4 10% 16% 93% Mobage 298 234.6 266.7 205.9 11% 12% 21% 268.7 205.2 10% 13% 100% Gree 387.8 249.6 296.8 173.5 23% 30% 68% 310.7 172.5 20% 31% 88%

§  Processed Image Ra9o = (Total no. of processed image / Total no. of origin image) §  We can know how much image data can be op9mized.

U9lizing Cache Server

origin Image Opt. Cache client

Service Flow

h[p://www.foo.com/a.png Cache-‐miss h[p://www.foo.com/a.png

h[p://www.foo.com/a.png

Content-‐Type : Image/PNG : a.png

Image convert : PNG -‐> JPEG

Content-‐Type : Image/JPEG : a.png

Cache-‐fill Content-‐Type : Image/JPEG : a.png

h[p://www.foo.com/a.png

Content-‐Type : Image/JPEG : a.png Cache-‐hit

§  Image op9miza9on is very CPU-‐intensive job. To alleviate load, it’s be[er to u9lize cache server between client and image op9miza9on.

Performance of FEO and Image Op9miza9on

INTERNET

Test Environment

Origin Web

Gomez Agent

KR

Cdnetworks Cache POP

Gomez Agent in CA

US 2. Accelera9on of dynamic content by CDNW DWA

Gomez Agent

UK Singapore

CDNW POP at Dallas, US

Origin loca9on : Dallas, US MIO & MOD loca9on : Dallas, US CDNW Shield loca9on : Chicago, US

MIO MOD

MIO : mod image op9mizer MOD : mod page speed

Performance Comparison

Gomez Agent origin 1. between gomez agent and origin

2. between gomez agent and MIO

3. between gomez agent and CDNW DWA

4. between gomez agent and CDNW DWA + MOD

5. between gomez agent and CDNW DWA + MIO

Web Contents for PC & Mobile

Test scenario Test case

CDNW EDGE CDNW SHIELD

MIO

MIO

MOD

Performance Test using Gomez PC Agent §  Gomez Agent loca9ons

§  All : San Jose, Atlanta, New York, Seoul, Tokyo, Madrid, London

§  Using Gomez FireFox Agent §  1. origin & 4.DWA

§  Total Objects : 195 §  Total Bytes : 1004KB

§  6. DWA+MOD §  Total Objects : 49 §  Total Bytes : 652KB

§  7. DWA+MIO §  Total Objects : 195 §  Total Bytes : 518KB

§  Gomez UA String §  Mozilla/5.0 (Windows NT 6.1; WOW64; rv:13.0

; GomezAgent 3.0) Gecko/20100101 Firefox/13.0.1

4 3 1 2

Performance Test using Gomez Mobile Agent §  Gomez Agent loca9ons

§  All: Tokyo, Seoul, London, New York, Santa Clara

§  Using Gomez Mobile Agent §  1. origin & 4. DWA

§  Total Objects : 41 §  Total Bytes : 323.3KB

§  6. DWA+MPS(mod_pagespeed) §  Total Objects : 28 §  Total Bytes : 188.4KB

§  7. DWA+MIO §  Total Objects : 41 §  Total Bytes : 146.5KB

§  Gomez UA String §  Mozilla/5.0 (Windows NT 6.1; WOW6

4; rv:13.0; GomezRecorder 5.0) Gecko/20100101 Firefox/13.0.1

4 3 1 2

Mobile CDN

Introduc9on §  Mobile operators needs :

§  Deliver be[er QoE to customers §  Network op9miza9on

§  Alleviate mobile traffic conges9on

§  Reduce traffic and costs of OTT Content §  Save interna9onal transit traffic

§  Generate new revenue

What is Mobile CDN §  Op9mize the delivery of content to end users on any type of wirele

ss or mobile network. §  Mobile CDN is important for MNOs (Mobile Network Operator) as t

hey can lead to significant savings and avoid network conges9on. §  Provide be[er UX (User Experience) and traffic reduc9on

Loca9on of CDN inside the mobile network §  To get op9mal reduc9on and accelera9on, the CDN is placed inside

the mobile network on top of the PGW func9on.

Internet

PGW

eNB eNB eNB eNB eNB eNB eNB eNB eNB

SGW SGW SGW

IP Core

LTE Core

Mail Web ・・・ traffic is reduced to 1/10

-‐  Video pacing

-‐  Text Compression

-‐  Image Op*miza*on

-‐  TCP Accelera*on

In case of LTE

NAT Cache Edge

inside

Mobile NW

* Network Address Translator

* Primary Gateway

* Secondary Gateway

Video Pacing §  Video pacing controls the bandwidth for progressive download video. §  This reduces excessive video download

With Video Pacing Without Video Pacing Burst sec*on

Media’s bitrate Smooth sec*on

Conclusion

Conclusion

QoE

[L4 improvement] TCP （Transport-‐level）Op*miza*on

[L5 improvement] Adopt SPDY/RE Technology HTTP (Session-‐level) op*miza*on

[L6 improvement] Adopt FEO Technology HTML (Presenta*on-‐level) Op*miza*on

[L7 improvement] Content Op*miza*on Video Pacing, Image Op*miza*on and etc.

[L1-‐L3 improvement] Adopt CDN (beZer performance than ISP N/W)

•  To reduce latency on mobile network, place content closer to end-‐user. To do so, u9lize CDN. Also can accelerate middle mile.

•  Tune TCP depending on network type (3G/LTE/Wired).

•  FEO provides be[er QoE. And tune FEO focused on display 9me, not network download 9me.

•  Op9mize image. As reducing image size, can reduce traffic and provide be[er QoE.

The more QoE is improved, the more revenue is increased