SlideShare ist ein Scribd-Unternehmen logo
1 von 9
Downloaden Sie, um offline zu lesen
Remote Display Protocol Performance,
Thin Client Computing, and IPQ:
Test Report

Tests conducted and report prepared by




1 February 2011
Remote Display Protocol Performance, Thin Client Computing, and IPQ:
Test Report
Contents


      Executive Summary
      Test Setup
      Test Description
      Test Data Collection and Evaluation Considerations
            Packet Loss
            TCP Retransmissions
            Quality of Experience MOS (Mean Opinion Score)
      Test Results
            Packet Loss
            TCP Retransmissions
            Quality of Experience MOS (Mean Opinion Score)




Report Prepared by                                                     Page 2 of 9
Remote Display Protocol Performance, Thin Client Computing and IPQ:
Test Report
Executive Summary


Executive Summary
A leading vendor of thin client computing devices engaged IPeak Networks to test the performance of
one of their devices running each of three standard and widely remote display protocols over a lossy
WAN protected by IPeak Networks’ patented QoS software called IPQ.

The test results show that IPQ improved the performance of those protocols in several important ways.
IPQ consistently achieved better than 90% reduction in network packet loss and a substantial mitigation
of the damaging effects of packet loss on the performance of PCoIP, ICA and RDP. On average, IPQ
reduced 1% packet loss to less than 0.08%, 3% packet loss was reduced to less than 0.16%, and 5% was
reduced to less than 0.25%.

IPQ also enabled an average reduction in the numbers of TCP retransmissions (where applicable) of
better than 87% across all test scenarios. The reduction in TCP retransmissions supported welcome
relief from the user experience of performance lag and delays in application responsiveness.

The improvements in the quality of experience were impressive. Remote sessions suffering 3% and 5%
packet loss were essentially unusable without IPQ protection but achieved high Mean Opinion Scores for
quality of experience when IPQ was enabled.




Report Prepared by                                                                          Page 3 of 9
Remote Display Protocol Performance, Thin Client Computing and IPQ:
Test Report
Details and Interpretation


Test Setup
A leading vendor of thin client computing devices engaged IPeak Networks to test the performance of
one of their devices running each of three standard and widely remote display protocols over a lossy
WAN protected by IPeak Networks’ patented QoS software called IPQ. The physical test system and
network setup consisted of an off-the-shelf thin client computing device running Windows XP
Embedded and connected through a Layer 2 switch to a VMWare ESX server located in the IPeak
Networks data center. IPQ protection was added to this system and network setup. At the user end,
IPQ protection was enabled by installing the IPQ 32 bit Windows client on the thin client computing
device. In the data center, an IPQ Virtual Appliance was deployed on the ESX server. In addition, a
virtual switch was configured on the ESX server to ensure that all traffic from the thin client computing
device to the Windows XP virtual machine (VM) was bridged through the IPQ Virtual Appliance, all as
shown in the diagram below.




This test network was a high-quality LAN supporting 100 Mbps links and less than 1ms of latency. To
simulate a typically lossy WAN, IPeak Networks took the following steps:
 bandwidth between the IPQ Virtual Appliance and the Windows XP Desktop was limited to 5 Mbps,
 30ms of latency was added in each direction, and
 varying amounts of packet loss were introduced.




Report Prepared by                                                                           Page 4 of 9
Remote Display Protocol Performance, Thin Client Computing, and IPQ:
Test Report
Details and Interpretation


Test Description
Script-based application testing was used to test PCoIP, ICA and RDP. The script ran through a suite of
four user activities that characterize a common virtual desktop computing use-case:
     entering data and graphing the results in an MS Excel spreadsheet,
     viewing a MS PowerPoint presentation,
     playing a short video on YouTube, and
     scrolling through a PDF file in Adobe Reader.

Tests were run with each of the three remote display protocols—first without and then with IPQ
protection enabled—while packet loss was introduced at 0%, 1%, 3% and 5% levels. The generally
available Linux TC network simulation tools were used to introduce the packet loss on the link from the
IPQ Virtual Appliance to the thin client computing device, the direction in which the bulk of data flows in
a virtual desktop deployment.

Test Data Collection and Evaluation Considerations

Packet Loss
Packet loss degrades network performance which in turn degrades application performance and the
user quality of experience. Packet loss data—total average network loss with and without IPQ
enabled—was collected using the logging facilities built in to the IPQ client installed on the thin client
computing device.

TCP Retransmissions
TCP retransmissions increase dramatically in the presence of network packet loss. TCP retransmissions
manifest as delays and input/response lag which degrade the user’s quality of experience of real time
network applications. Where applicable, TCP retransmission data was also captured by running
Wireshark at the Windows XP Virtual Desktop.

Mean Opinion Score (MOS)
Quality of Experience was evaluated was and a Mean Opinion Score (MOS) assigned based on the test
operators’ observations with each combination of protocol, loss rate, user activity, and IPQ status (on or
off). A total of ninety-six scores were recorded.

The IPeak Networks MOS scheme and evaluation criteria are as shown in the table below.

 MOS       Quality        Impairment
 5         Excellent      Imperceptible
 4         Good           Perceptible but not annoying
 3         Fair           Very apparent and annoying
 2         Poor           Very Annoying
 1         Bad            Unusable



Report Prepared by                                                                             Page 5 of 9
Remote Display Protocol Performance, Thin Client Computing, and IPQ:
Test Report
Details and Interpretation


Results

Packet Loss
With IPQ protection against packet loss enabled, the reduction in packet loss was significant, with all
three remote display protocols and at all levels of loss. The summarized results are shown in the
following table and graphs.

                                                 IPQ OFF                          IPQ ON
                                               Total Average    Total Average   Improvement   Percentage
   Protocol               Injected Loss
                                               Network Loss     Network Loss       Factor      Reduction
                               1%                 1.09%             0.06%          18.2          94%
         PCoIP                 3%                 3.18%             0.17%          18.7          95%
                               5%                 4.91%             0.25%          19.6          95%
                               1%                 1.27%             0.08%          15.9          94%
                 ICA           3%                 3.53%             0.08%          44.1          98%
                               5%                 5.58%             0.26%          21.5          95%
                               1%                 1.14%             0.10%          11.4          91%
            RDP                3%                 3.66%             0.23%          15.9          94%
                               5%                 5.97%             0.24%          24.9          96%




                                    PCoIP
                  5.00%
                  4.00%
   Actual Loss




                  3.00%
                  2.00%                                        IPQ OFF
                  1.00%                                        IPQ ON
                  0.00%
                          1%         3%         5%

                               Injected Loss




Report Prepared by                                                                              Page 6 of 9
Remote Display Protocol Performance, Thin Client Computing, and IPQ:
Test Report
Details and Interpretation




                                  RDP
                 6.00%
                 5.00%
   Actual Loss




                 4.00%
                 3.00%
                                                   IPQ OFF
                 2.00%
                 1.00%                             IPQ ON
                 0.00%
                         1%        3%         5%
                              Injected Loss




                                  ICA
                 6.00%
                 5.00%
   Actual Loss




                 4.00%
                 3.00%
                                                   IPQ OFF
                 2.00%
                                                   IPQ ON
                 1.00%
                 0.00%
                         1%        3%         5%

                              Injected Loss




Report Prepared by                                                     Page 7 of 9
Remote Display Protocol Performance, Thin Client Computing, and IPQ:
Test Report
Details and Interpretation


TCP Retransmissions
With IPQ enabled, the reduction in TCP retransmissions was significant. The summarized test results are
shown in the following table and graphs.

                                            IPQ OFF            IPQ ON
  Protocol                      Loss Level                         TCP Retrans
                                           TCP Retrans TCP Retrans
                                                                    Reduction
  PCoIP                         1%             NA          NA          NA
                                3%             NA          NA          NA
                                5%             NA          NA          NA
  ICA                           1%             >500        <150       70%
                                3%            >1700        <150       91%
                                5%            >3000        <250       95%
  RDP                           1%             >500        <100       80%
                                3%            >1900        <100       95%
                                5%            >4000        <250       94%



                          ICA and RDP TCP Retransmissions
                         5000
   TCP Retransmissions




                         4000

                         3000                                           RDP IPQ OFF
                         2000                                           RDP IPQ ON
                         1000                                           ICA IPQ OFF

                            0                                           "ICA IPQ ON"
                                   0%      1%        3%    5%
                                           Injected Loss




Note that this measurement is not applicable to PCoIP which is a UDP-based remote display protocol.




Report Prepared by                                                                          Page 8 of 9
Remote Display Protocol Performance, Thin Client Computing, and IPQ:
Test Report
Details and Interpretation


Mean Opinion Score (MOS)
The IPeak Networks MOS scheme and evaluation criteria are as shown in the table below.

 MOS        Quality        Impairment
 5          Excellent      Imperceptible
 4          Good           Perceptible but not annoying
 3          Fair           Very apparent and annoying
 2          Poor           Very Annoying
 1          Bad            Unusable



When network packet loss was present, the quality of experience was degraded very rapidly, regardless
of protocol. However, when IPQ was enabled, the quality of experience remained very consistent, even
when there was up to 5% loss on the network.

                                 IPQ Off                  IPQ ON

 Protocol     Loss Level     MOS      Quality    MOS          Quality

  PCoIP        1%            3.5        Good      4.75       Excellent
               3%            2.5        Fair      4.5        Excellent
               5%            2          Fair      4          Good
  ICA          1%            3.5        Good      5          Excellent
               3%            2.5        Fair      4.75       Excellent
               5%            1.5        Poor      4.25       Good
  RDP          1%            1.75       Poor      2.5        Fair
               3%            1          Poor      2          Fair
               5%            1          Poor      2          Fair




Report Prepared by                                                                        Page 9 of 9

Weitere ähnliche Inhalte

Was ist angesagt?

Netrounds Product Sheet
Netrounds Product SheetNetrounds Product Sheet
Netrounds Product SheetMarcus Friman
 
AirCheck
AirCheckAirCheck
AirChecklenlax
 
860 dspi voip_rtp_find_and_fix
860 dspi voip_rtp_find_and_fix860 dspi voip_rtp_find_and_fix
860 dspi voip_rtp_find_and_fixtrilithicweb
 
/conferences/spr2002/presentations/kirby.ppt
/conferences/spr2002/presentations/kirby.ppt/conferences/spr2002/presentations/kirby.ppt
/conferences/spr2002/presentations/kirby.pptVideoguy
 
Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...
Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...
Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...Veryx Technologies
 
Setup VoIP System and Interconnection with LTE network
Setup VoIP System and Interconnection with LTE networkSetup VoIP System and Interconnection with LTE network
Setup VoIP System and Interconnection with LTE networkNazmul Hossain Rakib
 
BGP Advance Technique by Steven & James
BGP Advance Technique by Steven & JamesBGP Advance Technique by Steven & James
BGP Advance Technique by Steven & JamesFebrian ‎
 
6 understanding aruba rf issues
6 understanding aruba rf issues6 understanding aruba rf issues
6 understanding aruba rf issuesVenudhanraj
 

Was ist angesagt? (20)

Netrounds Product Sheet
Netrounds Product SheetNetrounds Product Sheet
Netrounds Product Sheet
 
AirCheck
AirCheckAirCheck
AirCheck
 
EMEA Airheads - Aruba Remote Access Point (RAP) Troubleshooting
EMEA Airheads - Aruba Remote Access Point (RAP) TroubleshootingEMEA Airheads - Aruba Remote Access Point (RAP) Troubleshooting
EMEA Airheads - Aruba Remote Access Point (RAP) Troubleshooting
 
ClearPass 6.3.6 Release Notes
ClearPass 6.3.6 Release NotesClearPass 6.3.6 Release Notes
ClearPass 6.3.6 Release Notes
 
860 dspi voip_rtp_find_and_fix
860 dspi voip_rtp_find_and_fix860 dspi voip_rtp_find_and_fix
860 dspi voip_rtp_find_and_fix
 
Airheads Tech Talks: Cloud Guest SSID on Aruba Central
Airheads Tech Talks: Cloud Guest SSID on Aruba CentralAirheads Tech Talks: Cloud Guest SSID on Aruba Central
Airheads Tech Talks: Cloud Guest SSID on Aruba Central
 
EMEA Airheads- ArubaOS - Rogue AP troubleshooting
EMEA Airheads- ArubaOS - Rogue AP troubleshootingEMEA Airheads- ArubaOS - Rogue AP troubleshooting
EMEA Airheads- ArubaOS - Rogue AP troubleshooting
 
/conferences/spr2002/presentations/kirby.ppt
/conferences/spr2002/presentations/kirby.ppt/conferences/spr2002/presentations/kirby.ppt
/conferences/spr2002/presentations/kirby.ppt
 
EMEA Airheads- ArubaOS - Understanding Control-Plane-Security
EMEA Airheads-  ArubaOS - Understanding Control-Plane-SecurityEMEA Airheads-  ArubaOS - Understanding Control-Plane-Security
EMEA Airheads- ArubaOS - Understanding Control-Plane-Security
 
Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...
Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...
Webinar - Achieving ce 2.0 network integrity - a solid foundation to enable t...
 
EMEA Airheads- Instant AP- APP REF and Mixed IAP Cluster deployments
EMEA Airheads- Instant AP- APP REF and Mixed IAP Cluster deploymentsEMEA Airheads- Instant AP- APP REF and Mixed IAP Cluster deployments
EMEA Airheads- Instant AP- APP REF and Mixed IAP Cluster deployments
 
Setup VoIP System and Interconnection with LTE network
Setup VoIP System and Interconnection with LTE networkSetup VoIP System and Interconnection with LTE network
Setup VoIP System and Interconnection with LTE network
 
BGP Advance Technique by Steven & James
BGP Advance Technique by Steven & JamesBGP Advance Technique by Steven & James
BGP Advance Technique by Steven & James
 
6 understanding aruba rf issues
6 understanding aruba rf issues6 understanding aruba rf issues
6 understanding aruba rf issues
 
Best Practices on Migrating to 802.11ac Wi-Fi #AirheadsConf Italy
Best Practices on Migrating to 802.11ac Wi-Fi #AirheadsConf ItalyBest Practices on Migrating to 802.11ac Wi-Fi #AirheadsConf Italy
Best Practices on Migrating to 802.11ac Wi-Fi #AirheadsConf Italy
 
EMEA Airheads - AP Discovery Logic and AP Deployment
EMEA Airheads - AP Discovery Logic and AP DeploymentEMEA Airheads - AP Discovery Logic and AP Deployment
EMEA Airheads - AP Discovery Logic and AP Deployment
 
Breakout - Airheads Macau 2013 - Top 10 Tips from Aruba TAC
Breakout - Airheads Macau 2013 - Top 10 Tips from Aruba TAC Breakout - Airheads Macau 2013 - Top 10 Tips from Aruba TAC
Breakout - Airheads Macau 2013 - Top 10 Tips from Aruba TAC
 
EMEA Airheads- Instant AP- Instant AP Best Practice Configuration
EMEA Airheads- Instant AP- Instant AP Best Practice ConfigurationEMEA Airheads- Instant AP- Instant AP Best Practice Configuration
EMEA Airheads- Instant AP- Instant AP Best Practice Configuration
 
EMEA Airheads – Aruba controller features used to optimize performance
EMEA Airheads – Aruba controller features used to optimize performanceEMEA Airheads – Aruba controller features used to optimize performance
EMEA Airheads – Aruba controller features used to optimize performance
 
Design Fundamentals for Remote and Branch Access Networks
Design Fundamentals for Remote and Branch Access NetworksDesign Fundamentals for Remote and Branch Access Networks
Design Fundamentals for Remote and Branch Access Networks
 

Andere mochten auch (9)

Thin Client
Thin ClientThin Client
Thin Client
 
Chanthy.rath sna b-gkit
Chanthy.rath sna b-gkitChanthy.rath sna b-gkit
Chanthy.rath sna b-gkit
 
October 2008 - Synergizing The Generations Discussion
October 2008 - Synergizing The Generations DiscussionOctober 2008 - Synergizing The Generations Discussion
October 2008 - Synergizing The Generations Discussion
 
Thin Client Computing In The Steady State
Thin Client Computing In The Steady StateThin Client Computing In The Steady State
Thin Client Computing In The Steady State
 
Curso google adwords (português)
Curso google adwords (português)Curso google adwords (português)
Curso google adwords (português)
 
Thin client
Thin clientThin client
Thin client
 
ThinClient
ThinClientThinClient
ThinClient
 
Thin Client
Thin ClientThin Client
Thin Client
 
Thin client
Thin clientThin client
Thin client
 

Ähnlich wie Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report

Evaluating Quality of Experience in IPTV Services Using MPEG Frame Loss Rate
Evaluating Quality of Experience in IPTV Services Using MPEG Frame Loss RateEvaluating Quality of Experience in IPTV Services Using MPEG Frame Loss Rate
Evaluating Quality of Experience in IPTV Services Using MPEG Frame Loss RateJorge E. López de Vergara Méndez
 
PLNOG 13: Piotr Głaska: Quality of service monitoring in IP networks
PLNOG 13: Piotr Głaska: Quality of service monitoring in IP networksPLNOG 13: Piotr Głaska: Quality of service monitoring in IP networks
PLNOG 13: Piotr Głaska: Quality of service monitoring in IP networksPROIDEA
 
PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...
PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...
PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...PROIDEA
 
IPTV Service Assurance
IPTV Service AssuranceIPTV Service Assurance
IPTV Service AssuranceTTI Telecom
 
IPQ and RDP Test Report
IPQ and RDP Test ReportIPQ and RDP Test Report
IPQ and RDP Test ReportIPeak Networks
 
Advantage of IP system & Panasonic Security_Ver1.ppt
Advantage of IP system & Panasonic Security_Ver1.pptAdvantage of IP system & Panasonic Security_Ver1.ppt
Advantage of IP system & Panasonic Security_Ver1.pptPawachMetharattanara
 
2017_IMC_QUIC.pptx
2017_IMC_QUIC.pptx2017_IMC_QUIC.pptx
2017_IMC_QUIC.pptxBrian Zein
 
Electronics Division Linked In
Electronics Division   Linked InElectronics Division   Linked In
Electronics Division Linked Injpguy
 
Introduction to Cisco IPCC Enterprise
Introduction to Cisco IPCC EnterpriseIntroduction to Cisco IPCC Enterprise
Introduction to Cisco IPCC Enterprisemann_chaddha
 
Airspan: Network Densification using Outdoor and Indoor Small Cells
Airspan: Network Densification using Outdoor and Indoor Small CellsAirspan: Network Densification using Outdoor and Indoor Small Cells
Airspan: Network Densification using Outdoor and Indoor Small CellsSmall Cell Forum
 
C08 – Updated planning and commissioning guidelines for Profinet - Xaver Sch...
C08 – Updated planning and commissioning guidelines for Profinet -  Xaver Sch...C08 – Updated planning and commissioning guidelines for Profinet -  Xaver Sch...
C08 – Updated planning and commissioning guidelines for Profinet - Xaver Sch...PROFIBUS and PROFINET InternationaI - PI UK
 
Cluster DT sample report
Cluster DT sample reportCluster DT sample report
Cluster DT sample reportanil parmar
 
Using ai and automation to build resiliency into azure dev ops
Using ai and automation to build resiliency into azure dev opsUsing ai and automation to build resiliency into azure dev ops
Using ai and automation to build resiliency into azure dev opsRob Jahn
 
Measuring IPv6 ISP performance
Measuring IPv6 ISP performanceMeasuring IPv6 ISP performance
Measuring IPv6 ISP performanceAPNIC
 
Second Quarter 2004 Results En
Second Quarter 2004 Results EnSecond Quarter 2004 Results En
Second Quarter 2004 Results EnTIM RI
 
istio: service mesh for all
istio: service mesh for allistio: service mesh for all
istio: service mesh for allMandar Jog
 
Bluecoat Packeshaper Presentation
Bluecoat Packeshaper PresentationBluecoat Packeshaper Presentation
Bluecoat Packeshaper PresentationJerry Paul Acosta
 
Can We Multiplex ACKs without Harming the Performance of TCP?
Can We Multiplex ACKs without Harming the Performance of TCP?Can We Multiplex ACKs without Harming the Performance of TCP?
Can We Multiplex ACKs without Harming the Performance of TCP?Jose Saldana
 
Foundation Fieldbus - Control in the Field
Foundation Fieldbus - Control in the FieldFoundation Fieldbus - Control in the Field
Foundation Fieldbus - Control in the FieldJim Cahill
 

Ähnlich wie Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report (20)

Evaluating Quality of Experience in IPTV Services Using MPEG Frame Loss Rate
Evaluating Quality of Experience in IPTV Services Using MPEG Frame Loss RateEvaluating Quality of Experience in IPTV Services Using MPEG Frame Loss Rate
Evaluating Quality of Experience in IPTV Services Using MPEG Frame Loss Rate
 
PLNOG 13: Piotr Głaska: Quality of service monitoring in IP networks
PLNOG 13: Piotr Głaska: Quality of service monitoring in IP networksPLNOG 13: Piotr Głaska: Quality of service monitoring in IP networks
PLNOG 13: Piotr Głaska: Quality of service monitoring in IP networks
 
PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...
PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...
PLNOG15: VidMon - monitoring video signal quality in Service Provider IP netw...
 
IPTV Service Assurance
IPTV Service AssuranceIPTV Service Assurance
IPTV Service Assurance
 
IPQ and RDP Test Report
IPQ and RDP Test ReportIPQ and RDP Test Report
IPQ and RDP Test Report
 
Advantage of IP system & Panasonic Security_Ver1.ppt
Advantage of IP system & Panasonic Security_Ver1.pptAdvantage of IP system & Panasonic Security_Ver1.ppt
Advantage of IP system & Panasonic Security_Ver1.ppt
 
2017_IMC_QUIC.pptx
2017_IMC_QUIC.pptx2017_IMC_QUIC.pptx
2017_IMC_QUIC.pptx
 
Electronics Division Linked In
Electronics Division   Linked InElectronics Division   Linked In
Electronics Division Linked In
 
Introduction to Cisco IPCC Enterprise
Introduction to Cisco IPCC EnterpriseIntroduction to Cisco IPCC Enterprise
Introduction to Cisco IPCC Enterprise
 
Airspan: Network Densification using Outdoor and Indoor Small Cells
Airspan: Network Densification using Outdoor and Indoor Small CellsAirspan: Network Densification using Outdoor and Indoor Small Cells
Airspan: Network Densification using Outdoor and Indoor Small Cells
 
C08 – Updated planning and commissioning guidelines for Profinet - Xaver Sch...
C08 – Updated planning and commissioning guidelines for Profinet -  Xaver Sch...C08 – Updated planning and commissioning guidelines for Profinet -  Xaver Sch...
C08 – Updated planning and commissioning guidelines for Profinet - Xaver Sch...
 
Cluster DT sample report
Cluster DT sample reportCluster DT sample report
Cluster DT sample report
 
Using ai and automation to build resiliency into azure dev ops
Using ai and automation to build resiliency into azure dev opsUsing ai and automation to build resiliency into azure dev ops
Using ai and automation to build resiliency into azure dev ops
 
Measuring IPv6 ISP performance
Measuring IPv6 ISP performanceMeasuring IPv6 ISP performance
Measuring IPv6 ISP performance
 
Second Quarter 2004 Results En
Second Quarter 2004 Results EnSecond Quarter 2004 Results En
Second Quarter 2004 Results En
 
istio: service mesh for all
istio: service mesh for allistio: service mesh for all
istio: service mesh for all
 
Bluecoat Packeshaper Presentation
Bluecoat Packeshaper PresentationBluecoat Packeshaper Presentation
Bluecoat Packeshaper Presentation
 
Cma5000a gige
Cma5000a   gigeCma5000a   gige
Cma5000a gige
 
Can We Multiplex ACKs without Harming the Performance of TCP?
Can We Multiplex ACKs without Harming the Performance of TCP?Can We Multiplex ACKs without Harming the Performance of TCP?
Can We Multiplex ACKs without Harming the Performance of TCP?
 
Foundation Fieldbus - Control in the Field
Foundation Fieldbus - Control in the FieldFoundation Fieldbus - Control in the Field
Foundation Fieldbus - Control in the Field
 

Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report

  • 1. Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report Tests conducted and report prepared by 1 February 2011
  • 2. Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report Contents  Executive Summary  Test Setup  Test Description  Test Data Collection and Evaluation Considerations  Packet Loss  TCP Retransmissions  Quality of Experience MOS (Mean Opinion Score)  Test Results  Packet Loss  TCP Retransmissions  Quality of Experience MOS (Mean Opinion Score) Report Prepared by Page 2 of 9
  • 3. Remote Display Protocol Performance, Thin Client Computing and IPQ: Test Report Executive Summary Executive Summary A leading vendor of thin client computing devices engaged IPeak Networks to test the performance of one of their devices running each of three standard and widely remote display protocols over a lossy WAN protected by IPeak Networks’ patented QoS software called IPQ. The test results show that IPQ improved the performance of those protocols in several important ways. IPQ consistently achieved better than 90% reduction in network packet loss and a substantial mitigation of the damaging effects of packet loss on the performance of PCoIP, ICA and RDP. On average, IPQ reduced 1% packet loss to less than 0.08%, 3% packet loss was reduced to less than 0.16%, and 5% was reduced to less than 0.25%. IPQ also enabled an average reduction in the numbers of TCP retransmissions (where applicable) of better than 87% across all test scenarios. The reduction in TCP retransmissions supported welcome relief from the user experience of performance lag and delays in application responsiveness. The improvements in the quality of experience were impressive. Remote sessions suffering 3% and 5% packet loss were essentially unusable without IPQ protection but achieved high Mean Opinion Scores for quality of experience when IPQ was enabled. Report Prepared by Page 3 of 9
  • 4. Remote Display Protocol Performance, Thin Client Computing and IPQ: Test Report Details and Interpretation Test Setup A leading vendor of thin client computing devices engaged IPeak Networks to test the performance of one of their devices running each of three standard and widely remote display protocols over a lossy WAN protected by IPeak Networks’ patented QoS software called IPQ. The physical test system and network setup consisted of an off-the-shelf thin client computing device running Windows XP Embedded and connected through a Layer 2 switch to a VMWare ESX server located in the IPeak Networks data center. IPQ protection was added to this system and network setup. At the user end, IPQ protection was enabled by installing the IPQ 32 bit Windows client on the thin client computing device. In the data center, an IPQ Virtual Appliance was deployed on the ESX server. In addition, a virtual switch was configured on the ESX server to ensure that all traffic from the thin client computing device to the Windows XP virtual machine (VM) was bridged through the IPQ Virtual Appliance, all as shown in the diagram below. This test network was a high-quality LAN supporting 100 Mbps links and less than 1ms of latency. To simulate a typically lossy WAN, IPeak Networks took the following steps:  bandwidth between the IPQ Virtual Appliance and the Windows XP Desktop was limited to 5 Mbps,  30ms of latency was added in each direction, and  varying amounts of packet loss were introduced. Report Prepared by Page 4 of 9
  • 5. Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report Details and Interpretation Test Description Script-based application testing was used to test PCoIP, ICA and RDP. The script ran through a suite of four user activities that characterize a common virtual desktop computing use-case:  entering data and graphing the results in an MS Excel spreadsheet,  viewing a MS PowerPoint presentation,  playing a short video on YouTube, and  scrolling through a PDF file in Adobe Reader. Tests were run with each of the three remote display protocols—first without and then with IPQ protection enabled—while packet loss was introduced at 0%, 1%, 3% and 5% levels. The generally available Linux TC network simulation tools were used to introduce the packet loss on the link from the IPQ Virtual Appliance to the thin client computing device, the direction in which the bulk of data flows in a virtual desktop deployment. Test Data Collection and Evaluation Considerations Packet Loss Packet loss degrades network performance which in turn degrades application performance and the user quality of experience. Packet loss data—total average network loss with and without IPQ enabled—was collected using the logging facilities built in to the IPQ client installed on the thin client computing device. TCP Retransmissions TCP retransmissions increase dramatically in the presence of network packet loss. TCP retransmissions manifest as delays and input/response lag which degrade the user’s quality of experience of real time network applications. Where applicable, TCP retransmission data was also captured by running Wireshark at the Windows XP Virtual Desktop. Mean Opinion Score (MOS) Quality of Experience was evaluated was and a Mean Opinion Score (MOS) assigned based on the test operators’ observations with each combination of protocol, loss rate, user activity, and IPQ status (on or off). A total of ninety-six scores were recorded. The IPeak Networks MOS scheme and evaluation criteria are as shown in the table below. MOS Quality Impairment 5 Excellent Imperceptible 4 Good Perceptible but not annoying 3 Fair Very apparent and annoying 2 Poor Very Annoying 1 Bad Unusable Report Prepared by Page 5 of 9
  • 6. Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report Details and Interpretation Results Packet Loss With IPQ protection against packet loss enabled, the reduction in packet loss was significant, with all three remote display protocols and at all levels of loss. The summarized results are shown in the following table and graphs. IPQ OFF IPQ ON Total Average Total Average Improvement Percentage Protocol Injected Loss Network Loss Network Loss Factor Reduction 1% 1.09% 0.06% 18.2 94% PCoIP 3% 3.18% 0.17% 18.7 95% 5% 4.91% 0.25% 19.6 95% 1% 1.27% 0.08% 15.9 94% ICA 3% 3.53% 0.08% 44.1 98% 5% 5.58% 0.26% 21.5 95% 1% 1.14% 0.10% 11.4 91% RDP 3% 3.66% 0.23% 15.9 94% 5% 5.97% 0.24% 24.9 96% PCoIP 5.00% 4.00% Actual Loss 3.00% 2.00% IPQ OFF 1.00% IPQ ON 0.00% 1% 3% 5% Injected Loss Report Prepared by Page 6 of 9
  • 7. Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report Details and Interpretation RDP 6.00% 5.00% Actual Loss 4.00% 3.00% IPQ OFF 2.00% 1.00% IPQ ON 0.00% 1% 3% 5% Injected Loss ICA 6.00% 5.00% Actual Loss 4.00% 3.00% IPQ OFF 2.00% IPQ ON 1.00% 0.00% 1% 3% 5% Injected Loss Report Prepared by Page 7 of 9
  • 8. Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report Details and Interpretation TCP Retransmissions With IPQ enabled, the reduction in TCP retransmissions was significant. The summarized test results are shown in the following table and graphs. IPQ OFF IPQ ON Protocol Loss Level TCP Retrans TCP Retrans TCP Retrans Reduction PCoIP 1% NA NA NA 3% NA NA NA 5% NA NA NA ICA 1% >500 <150 70% 3% >1700 <150 91% 5% >3000 <250 95% RDP 1% >500 <100 80% 3% >1900 <100 95% 5% >4000 <250 94% ICA and RDP TCP Retransmissions 5000 TCP Retransmissions 4000 3000 RDP IPQ OFF 2000 RDP IPQ ON 1000 ICA IPQ OFF 0 "ICA IPQ ON" 0% 1% 3% 5% Injected Loss Note that this measurement is not applicable to PCoIP which is a UDP-based remote display protocol. Report Prepared by Page 8 of 9
  • 9. Remote Display Protocol Performance, Thin Client Computing, and IPQ: Test Report Details and Interpretation Mean Opinion Score (MOS) The IPeak Networks MOS scheme and evaluation criteria are as shown in the table below. MOS Quality Impairment 5 Excellent Imperceptible 4 Good Perceptible but not annoying 3 Fair Very apparent and annoying 2 Poor Very Annoying 1 Bad Unusable When network packet loss was present, the quality of experience was degraded very rapidly, regardless of protocol. However, when IPQ was enabled, the quality of experience remained very consistent, even when there was up to 5% loss on the network. IPQ Off IPQ ON Protocol Loss Level MOS Quality MOS Quality PCoIP 1% 3.5 Good 4.75 Excellent 3% 2.5 Fair 4.5 Excellent 5% 2 Fair 4 Good ICA 1% 3.5 Good 5 Excellent 3% 2.5 Fair 4.75 Excellent 5% 1.5 Poor 4.25 Good RDP 1% 1.75 Poor 2.5 Fair 3% 1 Poor 2 Fair 5% 1 Poor 2 Fair Report Prepared by Page 9 of 9