SlideShare ist ein Scribd-Unternehmen logo
1 von 1
Downloaden Sie, um offline zu lesen
Speed Matters for Google Web Search
Jake Brutlag
Google, Inc.
June 22, 2009
Abstract – Experiments demonstrate that increasing
web search latency 100 to 400 ms reduces the daily
number of searches per user by 0.2% to 0.6%. Fur-
thermore, users do fewer searches the longer they are
exposed. For longer delays, the loss of searches per-
sists for a time even after latency returns to previous
levels.
Google runs experiments on search traffic to understand
and improve the search experience. A series of such ex-
periments injected different types of server-side delay into
the search results page load in order to understand the
impact of latency on user behavior. In a given experiment,
one group of users experienced the delay, while a second
group served as the control. Across the experiments, the
type of delay, the magnitude of the delay, and experiment
duration (number of weeks users subject to delay) varied.
Speed as perceived by the end user is driven by multi-
ple factors. These experiments model only one of them:
additional server processing time. Figure 1 illustrates the
three types of delay tested; the type of delay distinguishes
when the delay occurs in the sequence of steps that con-
stitute a page load.
Figure 1: Injecting Server-Side Delay
In Figure 1, time flows from left to right and diago-
nal lines represent network communication between the
client and server. A user perceives load time as the du-
ration from start of navigation until page complete. The
page load may include optional activities (in blue), but
every search includes an HTTP request for the page
HTML, spanning the duration between GET and Last
Byte. Server processing commences upon receipt of the
HTTP GET. Google search leverages HTTP chunking,
which means the page header can be transmitted (and
perhaps rendered) even before the results are ready.
The different types of delay tested are:
pre-header pausing all server processing for X ms
immediately upon receipt of the HTTP GET
post-header pausing for X ms after transmitting the
page header but before any further server
processing is done
post-ads pausing X ms mid-stream sending the results,
after the sponsored links
Both the post-header and post-ads delay may be masked
by network conditions. That is, a slow or unreliable con-
nection may mitigate any delay in server processing from
the user’s perspective. All three types of delay may be
perceived differently by users due to the degree of partial
rendering on the page.
All other things being equal, more usage, as measured
by number of searches, reflects more satisfied users. Ta-
ble 1 gives the average daily searches per user over the
experiment duration for the experiment group relative to
the control group.
Table 1: Experiment Impact on Daily Searches Per User
Type of Delay Magnitude Duration Impact
Pre-header 50 ms 4 weeks —
Pre-header 100 ms 4 weeks −0.20%
Post-header 200 ms 6 weeks −0.29%
Post-header 400 ms 6 weeks −0.59%
Post-ads 200 ms 4 weeks −0.30%
Average impact over 4 or 6 weeks hides any trend over
time. By focusing on the subset of users who were part of
the experiment (or control group) from the beginning (as
identified by a browser cookie), one can determine if there
is such a trend. Figure 2 illustrates the trend for the two 6
week experiments.
Figure 2: Impact of Post-header Delays Over Time
dailysearchesperuserrelativetocontrol
wk1 wk2 wk3 wk4 wk5 wk6
−1%−0.8%−0.6%−0.4%−0.2%0%0.2%
200 ms delay
400 ms delay
actual
trend
The change over time can be quantified by dividing the
6 week duration in half. The impact on daily searches
per user during the second half is −0.36% and −0.74%
for the 200 ms and 400 ms delay respectively, compared
to −0.22% and −0.44% for the first half. Users do fewer
searches the longer they are exposed to delay.
Furthermore, observing these users for the 400 ms de-
lay after we stop subjecting them to the delay, the rate of
daily searches per user for the experiment is still −0.21%
relative to the control (averaged over the 5 weeks af-
ter removal of the delay). For longer delays, the loss of
searches persists for a time even after latency returns to
previous levels.

Weitere ähnliche Inhalte

Ähnlich wie Study speed matters for google web search

Web Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger Bartel
Web Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger BartelWeb Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger Bartel
Web Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger BartelHolger Bartel
 
Quick guide to plan and execute a load test
Quick guide to plan and execute a load testQuick guide to plan and execute a load test
Quick guide to plan and execute a load testduke.kalra
 
Web App Performance Optimization services in UAE
Web App Performance Optimization services  in UAEWeb App Performance Optimization services  in UAE
Web App Performance Optimization services in UAEasiyahanif9977
 
Oral Presentation
Oral PresentationOral Presentation
Oral PresentationVideoguy
 
Importance of System Performance Testing - Bahaa Al Zubaidi.pdf
Importance of System Performance Testing - Bahaa Al Zubaidi.pdfImportance of System Performance Testing - Bahaa Al Zubaidi.pdf
Importance of System Performance Testing - Bahaa Al Zubaidi.pdfBahaa Al Zubaidi
 
Sending the data already gathered from the client to the Server
Sending the data already gathered from the client to the ServerSending the data already gathered from the client to the Server
Sending the data already gathered from the client to the Serverhussam242
 
Agile methodology
Agile methodologyAgile methodology
Agile methodologybipulpwc
 
Srs (Software Requirement Specification Document)
Srs (Software Requirement Specification Document) Srs (Software Requirement Specification Document)
Srs (Software Requirement Specification Document) Fatima Qayyum
 
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES ijwscjournal
 
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES ijwscjournal
 
Online examination system Documentation
Online examination system DocumentationOnline examination system Documentation
Online examination system DocumentationLehlohonoloMakoti
 
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES ijwscjournal
 
The impact of google instant on paid search
The impact of google instant on paid searchThe impact of google instant on paid search
The impact of google instant on paid searchSearch Talk
 
A Survey on: Utilizing of Different Features in Web Behavior Prediction
A Survey on: Utilizing of Different Features in Web Behavior PredictionA Survey on: Utilizing of Different Features in Web Behavior Prediction
A Survey on: Utilizing of Different Features in Web Behavior PredictionEditor IJMTER
 
Dynamic Organization of User Historical Queries
Dynamic Organization of User Historical QueriesDynamic Organization of User Historical Queries
Dynamic Organization of User Historical QueriesIJMER
 

Ähnlich wie Study speed matters for google web search (20)

28791456 web-testing
28791456 web-testing28791456 web-testing
28791456 web-testing
 
Web Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger Bartel
Web Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger BartelWeb Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger Bartel
Web Performance in the Age of HTTP2 - Topconf Tallinn 2016 - Holger Bartel
 
leucht_http2
leucht_http2leucht_http2
leucht_http2
 
Browser Based Performance Testing and Tuning
Browser Based Performance Testing and TuningBrowser Based Performance Testing and Tuning
Browser Based Performance Testing and Tuning
 
Quick guide to plan and execute a load test
Quick guide to plan and execute a load testQuick guide to plan and execute a load test
Quick guide to plan and execute a load test
 
Web App Performance Optimization services in UAE
Web App Performance Optimization services  in UAEWeb App Performance Optimization services  in UAE
Web App Performance Optimization services in UAE
 
Oral Presentation
Oral PresentationOral Presentation
Oral Presentation
 
Importance of System Performance Testing - Bahaa Al Zubaidi.pdf
Importance of System Performance Testing - Bahaa Al Zubaidi.pdfImportance of System Performance Testing - Bahaa Al Zubaidi.pdf
Importance of System Performance Testing - Bahaa Al Zubaidi.pdf
 
Sending the data already gathered from the client to the Server
Sending the data already gathered from the client to the ServerSending the data already gathered from the client to the Server
Sending the data already gathered from the client to the Server
 
Agile methodology
Agile methodologyAgile methodology
Agile methodology
 
Effects of Web Page Contents on Load Time over the Internet
Effects of Web Page Contents on Load Time over the InternetEffects of Web Page Contents on Load Time over the Internet
Effects of Web Page Contents on Load Time over the Internet
 
Srs (Software Requirement Specification Document)
Srs (Software Requirement Specification Document) Srs (Software Requirement Specification Document)
Srs (Software Requirement Specification Document)
 
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
 
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
 
Online examination system Documentation
Online examination system DocumentationOnline examination system Documentation
Online examination system Documentation
 
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
AVAILABILITY METRICS: UNDER CONTROLLED ENVIRONMENTS FOR WEB SERVICES
 
The impact of google instant on paid search
The impact of google instant on paid searchThe impact of google instant on paid search
The impact of google instant on paid search
 
A Survey on: Utilizing of Different Features in Web Behavior Prediction
A Survey on: Utilizing of Different Features in Web Behavior PredictionA Survey on: Utilizing of Different Features in Web Behavior Prediction
A Survey on: Utilizing of Different Features in Web Behavior Prediction
 
Dynamic Organization of User Historical Queries
Dynamic Organization of User Historical QueriesDynamic Organization of User Historical Queries
Dynamic Organization of User Historical Queries
 
SDLC Model
SDLC  ModelSDLC  Model
SDLC Model
 

Mehr von Bitsytask

Lehman Brothers ALT-A Mortgage Docs, December 18, 2006
Lehman Brothers ALT-A Mortgage Docs, December 18, 2006Lehman Brothers ALT-A Mortgage Docs, December 18, 2006
Lehman Brothers ALT-A Mortgage Docs, December 18, 2006Bitsytask
 
BNC Subprime Mortgage Ratesheet 7-2006
BNC Subprime Mortgage Ratesheet 7-2006BNC Subprime Mortgage Ratesheet 7-2006
BNC Subprime Mortgage Ratesheet 7-2006Bitsytask
 
Impac libor option arm 2nd matrix
Impac libor option arm 2nd matrixImpac libor option arm 2nd matrix
Impac libor option arm 2nd matrixBitsytask
 
New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...
New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...
New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...Bitsytask
 
Countrywide Option Arm Loans (Negative Amortization) July 26 2006
Countrywide Option Arm Loans (Negative Amortization) July 26 2006Countrywide Option Arm Loans (Negative Amortization) July 26 2006
Countrywide Option Arm Loans (Negative Amortization) July 26 2006Bitsytask
 
Lehman Brothers ALT-A mortgage outline August 18 2006
Lehman Brothers ALT-A mortgage outline August 18 2006Lehman Brothers ALT-A mortgage outline August 18 2006
Lehman Brothers ALT-A mortgage outline August 18 2006Bitsytask
 
Credit Suisse sellers guide (secondary market) August 2006
Credit Suisse sellers guide (secondary market) August 2006Credit Suisse sellers guide (secondary market) August 2006
Credit Suisse sellers guide (secondary market) August 2006Bitsytask
 
GMAC Mortgage Underwriting Guidelines 9-11-2006
GMAC Mortgage Underwriting Guidelines 9-11-2006GMAC Mortgage Underwriting Guidelines 9-11-2006
GMAC Mortgage Underwriting Guidelines 9-11-2006Bitsytask
 
Operation Ajax Declassified PDF 7 of 9
Operation Ajax Declassified PDF 7 of 9Operation Ajax Declassified PDF 7 of 9
Operation Ajax Declassified PDF 7 of 9Bitsytask
 
Operation Ajax Declassified PDF 6 of 9
Operation Ajax Declassified PDF 6 of 9Operation Ajax Declassified PDF 6 of 9
Operation Ajax Declassified PDF 6 of 9Bitsytask
 
Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9Bitsytask
 
Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9Bitsytask
 
Operation Ajax Declassified PDF 3 of 9
Operation Ajax Declassified PDF 3 of 9Operation Ajax Declassified PDF 3 of 9
Operation Ajax Declassified PDF 3 of 9Bitsytask
 
Operation Ajax Declassified PDF 2 of 9
Operation Ajax Declassified PDF 2 of 9Operation Ajax Declassified PDF 2 of 9
Operation Ajax Declassified PDF 2 of 9Bitsytask
 
Operation Ajax Declassified PDF 1 of 9
Operation Ajax Declassified PDF 1 of 9Operation Ajax Declassified PDF 1 of 9
Operation Ajax Declassified PDF 1 of 9Bitsytask
 
Operation Ajax Declassified PDF Appendix E
Operation Ajax Declassified PDF Appendix EOperation Ajax Declassified PDF Appendix E
Operation Ajax Declassified PDF Appendix EBitsytask
 
Operation Ajax Declassified PDF Appendix D
Operation Ajax Declassified PDF Appendix DOperation Ajax Declassified PDF Appendix D
Operation Ajax Declassified PDF Appendix DBitsytask
 
Operation Ajax Declassified PDF Appendix B
Operation Ajax Declassified PDF Appendix BOperation Ajax Declassified PDF Appendix B
Operation Ajax Declassified PDF Appendix BBitsytask
 
Operation Ajax Declassified PDF 9 of 9
Operation Ajax Declassified PDF 9 of 9Operation Ajax Declassified PDF 9 of 9
Operation Ajax Declassified PDF 9 of 9Bitsytask
 
the crypto republic
the crypto republicthe crypto republic
the crypto republicBitsytask
 

Mehr von Bitsytask (20)

Lehman Brothers ALT-A Mortgage Docs, December 18, 2006
Lehman Brothers ALT-A Mortgage Docs, December 18, 2006Lehman Brothers ALT-A Mortgage Docs, December 18, 2006
Lehman Brothers ALT-A Mortgage Docs, December 18, 2006
 
BNC Subprime Mortgage Ratesheet 7-2006
BNC Subprime Mortgage Ratesheet 7-2006BNC Subprime Mortgage Ratesheet 7-2006
BNC Subprime Mortgage Ratesheet 7-2006
 
Impac libor option arm 2nd matrix
Impac libor option arm 2nd matrixImpac libor option arm 2nd matrix
Impac libor option arm 2nd matrix
 
New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...
New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...
New Century Subprime Mortgage Matrix (Stated Doc / 80%, 550 FICO, 50% DTI) 7-...
 
Countrywide Option Arm Loans (Negative Amortization) July 26 2006
Countrywide Option Arm Loans (Negative Amortization) July 26 2006Countrywide Option Arm Loans (Negative Amortization) July 26 2006
Countrywide Option Arm Loans (Negative Amortization) July 26 2006
 
Lehman Brothers ALT-A mortgage outline August 18 2006
Lehman Brothers ALT-A mortgage outline August 18 2006Lehman Brothers ALT-A mortgage outline August 18 2006
Lehman Brothers ALT-A mortgage outline August 18 2006
 
Credit Suisse sellers guide (secondary market) August 2006
Credit Suisse sellers guide (secondary market) August 2006Credit Suisse sellers guide (secondary market) August 2006
Credit Suisse sellers guide (secondary market) August 2006
 
GMAC Mortgage Underwriting Guidelines 9-11-2006
GMAC Mortgage Underwriting Guidelines 9-11-2006GMAC Mortgage Underwriting Guidelines 9-11-2006
GMAC Mortgage Underwriting Guidelines 9-11-2006
 
Operation Ajax Declassified PDF 7 of 9
Operation Ajax Declassified PDF 7 of 9Operation Ajax Declassified PDF 7 of 9
Operation Ajax Declassified PDF 7 of 9
 
Operation Ajax Declassified PDF 6 of 9
Operation Ajax Declassified PDF 6 of 9Operation Ajax Declassified PDF 6 of 9
Operation Ajax Declassified PDF 6 of 9
 
Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9
 
Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9Operation Ajax Declassified PDF 5 of 9
Operation Ajax Declassified PDF 5 of 9
 
Operation Ajax Declassified PDF 3 of 9
Operation Ajax Declassified PDF 3 of 9Operation Ajax Declassified PDF 3 of 9
Operation Ajax Declassified PDF 3 of 9
 
Operation Ajax Declassified PDF 2 of 9
Operation Ajax Declassified PDF 2 of 9Operation Ajax Declassified PDF 2 of 9
Operation Ajax Declassified PDF 2 of 9
 
Operation Ajax Declassified PDF 1 of 9
Operation Ajax Declassified PDF 1 of 9Operation Ajax Declassified PDF 1 of 9
Operation Ajax Declassified PDF 1 of 9
 
Operation Ajax Declassified PDF Appendix E
Operation Ajax Declassified PDF Appendix EOperation Ajax Declassified PDF Appendix E
Operation Ajax Declassified PDF Appendix E
 
Operation Ajax Declassified PDF Appendix D
Operation Ajax Declassified PDF Appendix DOperation Ajax Declassified PDF Appendix D
Operation Ajax Declassified PDF Appendix D
 
Operation Ajax Declassified PDF Appendix B
Operation Ajax Declassified PDF Appendix BOperation Ajax Declassified PDF Appendix B
Operation Ajax Declassified PDF Appendix B
 
Operation Ajax Declassified PDF 9 of 9
Operation Ajax Declassified PDF 9 of 9Operation Ajax Declassified PDF 9 of 9
Operation Ajax Declassified PDF 9 of 9
 
the crypto republic
the crypto republicthe crypto republic
the crypto republic
 

Kürzlich hochgeladen

Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdfSimplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdfFIDO Alliance
 
Tales from a Passkey Provider Progress from Awareness to Implementation.pptx
Tales from a Passkey Provider  Progress from Awareness to Implementation.pptxTales from a Passkey Provider  Progress from Awareness to Implementation.pptx
Tales from a Passkey Provider Progress from Awareness to Implementation.pptxFIDO Alliance
 
Using IESVE for Room Loads Analysis - UK & Ireland
Using IESVE for Room Loads Analysis - UK & IrelandUsing IESVE for Room Loads Analysis - UK & Ireland
Using IESVE for Room Loads Analysis - UK & IrelandIES VE
 
Working together SRE & Platform Engineering
Working together SRE & Platform EngineeringWorking together SRE & Platform Engineering
Working together SRE & Platform EngineeringMarcus Vechiato
 
2024 May Patch Tuesday
2024 May Patch Tuesday2024 May Patch Tuesday
2024 May Patch TuesdayIvanti
 
Introduction to FIDO Authentication and Passkeys.pptx
Introduction to FIDO Authentication and Passkeys.pptxIntroduction to FIDO Authentication and Passkeys.pptx
Introduction to FIDO Authentication and Passkeys.pptxFIDO Alliance
 
Introduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdf
Introduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdfIntroduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdf
Introduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdfFIDO Alliance
 
Long journey of Ruby Standard library at RubyKaigi 2024
Long journey of Ruby Standard library at RubyKaigi 2024Long journey of Ruby Standard library at RubyKaigi 2024
Long journey of Ruby Standard library at RubyKaigi 2024Hiroshi SHIBATA
 
Portal Kombat : extension du réseau de propagande russe
Portal Kombat : extension du réseau de propagande russePortal Kombat : extension du réseau de propagande russe
Portal Kombat : extension du réseau de propagande russe中 央社
 
Event-Driven Architecture Masterclass: Challenges in Stream Processing
Event-Driven Architecture Masterclass: Challenges in Stream ProcessingEvent-Driven Architecture Masterclass: Challenges in Stream Processing
Event-Driven Architecture Masterclass: Challenges in Stream ProcessingScyllaDB
 
Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...
Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...
Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...ScyllaDB
 
Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...
Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...
Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...Skynet Technologies
 
WebAssembly is Key to Better LLM Performance
WebAssembly is Key to Better LLM PerformanceWebAssembly is Key to Better LLM Performance
WebAssembly is Key to Better LLM PerformanceSamy Fodil
 
The Metaverse: Are We There Yet?
The  Metaverse:    Are   We  There  Yet?The  Metaverse:    Are   We  There  Yet?
The Metaverse: Are We There Yet?Mark Billinghurst
 
How we scaled to 80K users by doing nothing!.pdf
How we scaled to 80K users by doing nothing!.pdfHow we scaled to 80K users by doing nothing!.pdf
How we scaled to 80K users by doing nothing!.pdfSrushith Repakula
 
Hyatt driving innovation and exceptional customer experiences with FIDO passw...
Hyatt driving innovation and exceptional customer experiences with FIDO passw...Hyatt driving innovation and exceptional customer experiences with FIDO passw...
Hyatt driving innovation and exceptional customer experiences with FIDO passw...FIDO Alliance
 
Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024Patrick Viafore
 
Breaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdfBreaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdfUK Journal
 
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdfThe Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdfFIDO Alliance
 

Kürzlich hochgeladen (20)

Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdfSimplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
 
Tales from a Passkey Provider Progress from Awareness to Implementation.pptx
Tales from a Passkey Provider  Progress from Awareness to Implementation.pptxTales from a Passkey Provider  Progress from Awareness to Implementation.pptx
Tales from a Passkey Provider Progress from Awareness to Implementation.pptx
 
Using IESVE for Room Loads Analysis - UK & Ireland
Using IESVE for Room Loads Analysis - UK & IrelandUsing IESVE for Room Loads Analysis - UK & Ireland
Using IESVE for Room Loads Analysis - UK & Ireland
 
Working together SRE & Platform Engineering
Working together SRE & Platform EngineeringWorking together SRE & Platform Engineering
Working together SRE & Platform Engineering
 
2024 May Patch Tuesday
2024 May Patch Tuesday2024 May Patch Tuesday
2024 May Patch Tuesday
 
Introduction to FIDO Authentication and Passkeys.pptx
Introduction to FIDO Authentication and Passkeys.pptxIntroduction to FIDO Authentication and Passkeys.pptx
Introduction to FIDO Authentication and Passkeys.pptx
 
Introduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdf
Introduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdfIntroduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdf
Introduction to FDO and How It works Applications _ Richard at FIDO Alliance.pdf
 
Long journey of Ruby Standard library at RubyKaigi 2024
Long journey of Ruby Standard library at RubyKaigi 2024Long journey of Ruby Standard library at RubyKaigi 2024
Long journey of Ruby Standard library at RubyKaigi 2024
 
Portal Kombat : extension du réseau de propagande russe
Portal Kombat : extension du réseau de propagande russePortal Kombat : extension du réseau de propagande russe
Portal Kombat : extension du réseau de propagande russe
 
Event-Driven Architecture Masterclass: Challenges in Stream Processing
Event-Driven Architecture Masterclass: Challenges in Stream ProcessingEvent-Driven Architecture Masterclass: Challenges in Stream Processing
Event-Driven Architecture Masterclass: Challenges in Stream Processing
 
Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...
Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...
Event-Driven Architecture Masterclass: Engineering a Robust, High-performance...
 
Overview of Hyperledger Foundation
Overview of Hyperledger FoundationOverview of Hyperledger Foundation
Overview of Hyperledger Foundation
 
Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...
Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...
Human Expert Website Manual WCAG 2.0 2.1 2.2 Audit - Digital Accessibility Au...
 
WebAssembly is Key to Better LLM Performance
WebAssembly is Key to Better LLM PerformanceWebAssembly is Key to Better LLM Performance
WebAssembly is Key to Better LLM Performance
 
The Metaverse: Are We There Yet?
The  Metaverse:    Are   We  There  Yet?The  Metaverse:    Are   We  There  Yet?
The Metaverse: Are We There Yet?
 
How we scaled to 80K users by doing nothing!.pdf
How we scaled to 80K users by doing nothing!.pdfHow we scaled to 80K users by doing nothing!.pdf
How we scaled to 80K users by doing nothing!.pdf
 
Hyatt driving innovation and exceptional customer experiences with FIDO passw...
Hyatt driving innovation and exceptional customer experiences with FIDO passw...Hyatt driving innovation and exceptional customer experiences with FIDO passw...
Hyatt driving innovation and exceptional customer experiences with FIDO passw...
 
Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024
 
Breaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdfBreaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdf
 
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdfThe Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
 

Study speed matters for google web search

  • 1. Speed Matters for Google Web Search Jake Brutlag Google, Inc. June 22, 2009 Abstract – Experiments demonstrate that increasing web search latency 100 to 400 ms reduces the daily number of searches per user by 0.2% to 0.6%. Fur- thermore, users do fewer searches the longer they are exposed. For longer delays, the loss of searches per- sists for a time even after latency returns to previous levels. Google runs experiments on search traffic to understand and improve the search experience. A series of such ex- periments injected different types of server-side delay into the search results page load in order to understand the impact of latency on user behavior. In a given experiment, one group of users experienced the delay, while a second group served as the control. Across the experiments, the type of delay, the magnitude of the delay, and experiment duration (number of weeks users subject to delay) varied. Speed as perceived by the end user is driven by multi- ple factors. These experiments model only one of them: additional server processing time. Figure 1 illustrates the three types of delay tested; the type of delay distinguishes when the delay occurs in the sequence of steps that con- stitute a page load. Figure 1: Injecting Server-Side Delay In Figure 1, time flows from left to right and diago- nal lines represent network communication between the client and server. A user perceives load time as the du- ration from start of navigation until page complete. The page load may include optional activities (in blue), but every search includes an HTTP request for the page HTML, spanning the duration between GET and Last Byte. Server processing commences upon receipt of the HTTP GET. Google search leverages HTTP chunking, which means the page header can be transmitted (and perhaps rendered) even before the results are ready. The different types of delay tested are: pre-header pausing all server processing for X ms immediately upon receipt of the HTTP GET post-header pausing for X ms after transmitting the page header but before any further server processing is done post-ads pausing X ms mid-stream sending the results, after the sponsored links Both the post-header and post-ads delay may be masked by network conditions. That is, a slow or unreliable con- nection may mitigate any delay in server processing from the user’s perspective. All three types of delay may be perceived differently by users due to the degree of partial rendering on the page. All other things being equal, more usage, as measured by number of searches, reflects more satisfied users. Ta- ble 1 gives the average daily searches per user over the experiment duration for the experiment group relative to the control group. Table 1: Experiment Impact on Daily Searches Per User Type of Delay Magnitude Duration Impact Pre-header 50 ms 4 weeks — Pre-header 100 ms 4 weeks −0.20% Post-header 200 ms 6 weeks −0.29% Post-header 400 ms 6 weeks −0.59% Post-ads 200 ms 4 weeks −0.30% Average impact over 4 or 6 weeks hides any trend over time. By focusing on the subset of users who were part of the experiment (or control group) from the beginning (as identified by a browser cookie), one can determine if there is such a trend. Figure 2 illustrates the trend for the two 6 week experiments. Figure 2: Impact of Post-header Delays Over Time dailysearchesperuserrelativetocontrol wk1 wk2 wk3 wk4 wk5 wk6 −1%−0.8%−0.6%−0.4%−0.2%0%0.2% 200 ms delay 400 ms delay actual trend The change over time can be quantified by dividing the 6 week duration in half. The impact on daily searches per user during the second half is −0.36% and −0.74% for the 200 ms and 400 ms delay respectively, compared to −0.22% and −0.44% for the first half. Users do fewer searches the longer they are exposed to delay. Furthermore, observing these users for the 400 ms de- lay after we stop subjecting them to the delay, the rate of daily searches per user for the experiment is still −0.21% relative to the control (averaged over the 5 weeks af- ter removal of the delay). For longer delays, the loss of searches persists for a time even after latency returns to previous levels.