Citrix vient de publier une note sur les « Best Practice » lync sous l’environnement Citrix en voici le contenu
This article contains information on Using Microsoft Lync 2010 with XenDesktop 5.5.
Background
The purpose of these tests is to assess the usability of Microsoft Lync 2010 Citrix XenDesktop 5.5 users. The tests were mainly carried out in the System 3 lab in Chalfont, but because this lab is isolated from the Citrix corporate LAN, it was necessary to use local Exchange and Lync servers, rather than Citrix corporate servers.Because the System 3 lab is independently connected to the Internet, the Citrix Technical Support Engineers were able to include various LAN/WAN topologies during testing. The engineers were also able to simulate LAN/WAN bandwidth, latency and packet loss conditions on our networks.
System 3 Lab – Hardware Details
- XenServer pool – 2 x Dell T7500 (Intel Xeon E5620 – 8 cores)
- Citrix Client PC (Eight) – Dell Precision T3500 – Windows 7 x32 – Logitech C600 webcam
-
Lync client PC (Nine) – Dell Precision 380 – Windows XP – Logitech C600 webcam
Software Details
- XenServer 5.6 Service Pack 2
- XenDesktop 5 (Alsace build 51153)
- Microsoft Lync 2010 (Client version 4.0.7577.275)
- Desktop Controller – Windows Server 2008 R2 Service Pack 1
- Virtual Desktop Agent – Windows 7 – 2 Gb memory – 2 virtual CPUs
XenDesktop Active Policy Settings
Network Details
The System 3 lab in Chalfont is connected to the Internet through a domestic ADSL service with a maximum upload speed of 1Mbps and a maximum download speed of 8Mbps. Because this ADSL service has multiple IP addresses, the engineers have configured our lab so that it is logically two separate sites:
- Virtdom
- Physdom
Although the engineers have no control over the speed of our ADSL line, network quality between Virtdom and Physdom sites is entirely under our control, and engineers use a Linktropy-mini WAN Emulator to simulate various LAN/WAN bandwidth, latency and packet loss conditions.
- In the following tables, latency is the total round trip time across the emulated network; for example, latency of 20Ms indicates that the link had 10Ms delay one way and 10Ms the other.
- During many of the tests some activity was generally performed on the desktop, such as moving windows, and typing into a text document a) in order to get an indication of desktop responsiveness, and b) to check for any effect that using the desktop might have on audio quality.
-
Due to the repetitive nature of this group of tests, it was generally not appropriate to hold live conversations with a person at each end of a call. For this reason, the primary (audio/visual source) user was generally replaced by an MP3 player or a tablet PC that was used to play media that included:
- Audio only – A BBC Podcast, such as The Strand or Great Lives, which are primarily speech rather than music.
- Audio and video – A YouTube video of a person reading from a book http://www.youtube.com/watch?v=Sxe6WxAVFqM. This video was chosen because it is a good example of a talking head.
- Audio only – A BBC Podcast, such as The Strand or Great Lives, which are primarily speech rather than music.
Initially when the engineers tried using the BBC news channel as an audio and video source but found that there was a so much screen activity that video stream (across the network) quickly fell behind the sound stream, causing an increasing lack of lip-sync. On reflection, the BBC news channel includes live news footage, as well as multiple camera angles of multiple talking heads. This is not what Lync was designed for.
-
Audio assessment
- Initial assessment of audio quality concentrated on the sound, and was based on factors such as a) were there any pops or breaks in the audio, and b) was the speech distorted in a way? As might be expected, as LAN/WAN quality was reduced, one eventually became aware that the audio quality was probably not be as good as in previous test runs. For assessment of this nature, it was useful to have a standard voice recording that could be repeated for each test.
- Although the preceding assessment of quality is useful, it was felt that for audio conferencing there were additional measures of usability, based on factors, such as a) am I getting the message, b) is the content understandable, and c) is audio quality a distraction. For assessments of this nature, it was much more useful to listen to a new Podcast. Generally, the engineers found that they had a much greater tolerance of poor network quality when assessing usability, rather than assessing audio quality.
- Initial assessment of audio quality concentrated on the sound, and was based on factors such as a) were there any pops or breaks in the audio, and b) was the speech distorted in a way? As might be expected, as LAN/WAN quality was reduced, one eventually became aware that the audio quality was probably not be as good as in previous test runs. For assessment of this nature, it was useful to have a standard voice recording that could be repeated for each test.
-
Audio and video assessment
- Generally, the assessments for audio quality and usability (above) are all applicable to audio in this section.
- For video, the problems encountered related to a) stuttering, or short freezes, and b) synchronisation of the picture and the sound (lip-sync). Of these, lack of lip-sync was the most common.
- Generally, the assessments for audio quality and usability (above) are all applicable to audio in this section.
Notes:
- The Audio conferencing software that was used was Microsoft Lync 2010
- The sound source used in these tests was an MP3 player, connected through a 3.5mm audio lead into the microphone port on the Citrix client PC (Eight). The MP3 player was playing BBC podcasts, such as The Strand and Great Lives.
- Generally, sharing of a desktop caused additional desktop lag at all network settings. However, there was little additional degradation of usability, probably because the additional lag was between the Lync parties, rather than between the Citrix client and the XenDesktop VDA.
Bandwidth set: 100 Mbps ( Linktropy max speed)- Audio Results
Total Latency (RTT)
Packet Loss Set
0%
0.5%
1%
20ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
40ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
80ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
150ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
200ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
500ms
Good Audio Quality
Desktop – some lag.
Good Audio Quality
Desktop – some lag.
Fair Audio Quality – Usable.
Desktop – some lag.
1000ms
Good Audio Quality
Desktop – some lag.
Good Audio Quality
Desktop – some lag.
Fair Audio Quality – Usable.
Desktop – significant lag.
2000ms
Good Audio Quality.
Desktop – significant lag.
Fair Audio Quality – Usable.
Desktop – significant lag.
Fair Audio Quality – Usable.
Desktop – Unresponsive.
Bandwidth set: 5 Mbps – Audio Results
Total Latency (RTT)
Packet Loss Set
0%
0.5%
1%
20ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
40ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
80ms
Good Audio Quality
Good Audio Quality
Fair Audio Quality – Usable.
Desktop – some lag.
150ms
Good Audio Quality
Good Audio Quality
Fair Audio Quality – Usable.
Desktop – some lag.
200ms
Good Audio Quality
Fair Audio Quality – Usable.
Desktop – some lag.
Fair Audio Quality – Usable.
Desktop – some lag.
500ms
Good Audio Quality
Desktop – some lag.
Fair Audio Quality – Usable.
Desktop – some lag.
Fair Audio Quality – Usable.
Desktop – significant lag.
1000ms
Good Audio Quality
Desktop – some lag.
Fair Audio Quality – Usable.
Desktop – significant lag.
Fair Audio Quality. Noticeable popping -Usable.
Desktop – Unresponsive.
2000ms
Good Audio Quality
Desktop – significant lag.
Fair Audio Quality – Usable.
Desktop – Unresponsive.
Fair Audio Quality. Noticeable popping -Usable.
Desktop – Unresponsive.
Audio Tests over WAN with Access Gateway and Branch Repeater
Notes:
- The Audio conferencing software that was used was Microsoft Lync 2010
- The sound source used in these tests was an MP3 player, connected through a 3.5mm audio lead into the microphone port on the Citrix client PC (Eight). The MP3 player was playing BBC podcasts, such as The Strand and Great Lives.
- Generally sharing of a desktop caused additional desktop lag at all network settings. However, there was slight additional degradation of usability, probably because the additional lag was between the Lync parties, rather than between the Citrix client and the XenDesktop VDA.
Bandwidth set: 5 Mbps Audio Results
Total Latency (RTT)
Packet Loss Set
0%
0.5%
1%
20ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
40ms
Good Audio Quality
Good Audio Quality
Good Audio Quality
80ms
Good Audio Quality
Good Audio Quality
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
150ms
Good Audio Quality
Good Audio Quality
Fair audio quality. Some popping and breakup especially with desktop activity.
Desktop – significant lag.
Probably usable
200ms
Good Audio Quality
Fair Audio Quality- usable
Desktop – Some lag
Fair audio quality. Some popping and breakup especially with desktop activity.
Desktop – significant lag.
Probably usable
500ms
Fair Audio Quality – Usable
Desktop – some lag.
Fair to poor audio Quality –
Noticeable popping and breakup
Desktop – Significant lag
Not usable.
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
1000ms
Fair audio quality. Some popping and breakup.
Desktop – some lag.
Usable
Not tested
Not tested
2000ms
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
Not tested
Not tested
Bandwidth set: 1.5 Mbps – Audio Results
Latency Set Eachway
Packet Loss Set
0%
0.5%
1%
20ms
Good audio quality
Good audio quality
Good audio quality
40ms
Good audio quality
Good audio quality
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
80ms
Good audio quality
Good audio quality
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
150ms
Good audio quality
Good audio quality
Fair audio quality. Some popping and breakup especially with desktop activity.
Desktop – significant lag.
Probably usable
200ms
Good audio quality
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
Fair audio quality. Some popping and breakup especially with desktop activity.
Desktop – significant lag.
Probably usable
500ms
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Probably not usable
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
1000ms
Fair audio quality. Some popping and breakup.
Desktop – some lag.
Usable
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
Not tested
2000ms
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
Not tested
Not tested
Bandwidth set: 768 kbps – Audio Results
Latency Set Eachway
Packet Loss Set
0%
0.5%
1%
20ms
Good audio quality
Good audio quality
Good audio quality
40ms
Good audio quality
Good audio quality
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
80ms
Good audio quality
Good audio quality
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
150ms
Good audio quality
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
Fair audio quality. Some popping and breakup especially with desktop activity.
Desktop – significant lag.
Probably usable
200ms
Good audio quality
Fair audio quality. Some popping and breakup especially with desktop activity.
Desktop – significant lag.
Probably usable
Fair audio quality. Some popping and breakup especially with desktop activity.
Desktop – significant lag.
Probably not usable
500ms
Fair audio quality. Slight popping and breakup.
Desktop – some lag.
Usable
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
1000ms
Moderate audio quality. Some popping and breakup.
Desktop – significant lag.
Probably usable
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
Not tested
2000ms
Poor audio quality. Significant popping and breakup.
Desktop – significant lag.
Not usable
Not tested
Not tested
Note:
The audio visual source used in these tests was a Logitech C600 webcam. This webcam was pointed at the tablet PC, and the audio-out from the tablet PC was connected through a 3.5mm audio lead into the microphone port on the Citrix client PC (Eight). The tablet PC was used to play http://www.youtube.com/watch?v=Sxe6WxAVFqM
Bandwidth set: 5 Mbps Audio & Video Results |
|||
Total Latency (RTT) |
Packet Loss Set |
||
0% |
0.5% |
1% |
|
20ms |
Sound quality – Good Video quality – Good Talking head A/V sync – Good |
Sound quality – Fair. Some popping Video quality – Good Talking head A/V sync – Good Usable |
Sound quality – Fair. Some popping Video quality – Good Talking head A/V sync – Good Usable |
40ms |
Sound quality – Good Video quality – Good Talking head A/V sync – Good |
Sound quality – Fair. Some popping Video quality – Good Talking head A/V sync – Fair Usable |
Sound quality – Fair. Some popping Video quality – Good Talking head A/V sync – Good Usable |
80ms |
Sound quality – Good Video quality – Good Talking head A/V sync – Good |
Sound quality – Fair. Some popping Video quality – Good Talking head A/V sync – Fair Usable |
Sound quality – Fair. Some popping Video quality – Good Talking head A/V sync – Fair Usable |
150ms |
Sound quality – Good Video quality – Good Talking head A/V sync – Good |
Sound quality – Fair. Some popping Video quality – Good Talking head A/V sync – Fair Usable |
Sound quality – Fair. Some popping Video quality – Fair Talking head A/V sync – Poor Desktop – Some lag Usable |
200ms |
Sound quality – Good Video quality – Good Talking head A/V sync – Good |
Sound quality – Fair. Some popping Video quality – Fair Talking head A/V sync – Fair Usable |
Sound quality – Fair. Some popping Video quality – Fair Talking head A/V sync – Poor Desktop – Significant lag Possibly usable |
500ms |
Sound quality – Fair Video quality – Fair Talking head A/V sync – Fair Desktop – Significant lag Usable |
Sound quality – Fair Video quality – Poor Talking head A/V sync – Poor Desktop – Significant lag Probably not usable |
Sound quality – Fair. Some popping Video quality – Poor Talking head A/V sync – Poor Desktop – Unresponsive Probably not usable |
1000ms |
Sound quality – Fair Video quality – Fair Talking head A/V sync – Fair Desktop – Significant lag Usable |
Sound quality – Fair Video quality – Poor Talking head A/V sync – Poor Desktop – Unresponsive Not usable |
Sound quality – Fair. Some popping Video quality – Poor Talking head A/V sync – Poor Desktop – Unresponsive Not usable |
2000ms |
Sound quality – Poor Video quality – Poor Talking head A/V sync – Poor Desktop – Unresponsive Not usable |
Not tested |
Not tested |
Audio/Video Tests over WAN with Access Gateway and Branch Repeater
Notes:
- The audio visual source used in these tests was a Logitech C600 webcam. This webcam was pointed at the tablet PC, and the audio-out from the tablet PC was connected through a 3.5mm audio lead into the microphone port on the Citrix client PC (Eight). The tablet PC was used to play http://www.youtube.com/watch?v=Sxe6WxAVFqM
- With regard to A/V lip-sync across a WAN, it is noticeable that lip-sync tends to drift out over time, and once out it rarely regains synchronisation. This is in contrast with the same tests when run in a LAN environment. In a LAN environment, and where UDP based audio is being used, the tendency is for lip-sync to be re-established during periods of low screen activity
Bandwidth set: 5 Mbps Audio & Video Results
Total Latency (RTT)
Packet Loss Set
0%
0.5%
1%
20ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Sound quality – Good
Video quality – Fair
Talking head A/V sync – Poor
Probably usable
Sound quality – Fair
Video quality – Fair
Talking head A/V sync – Poor.
Probably not usable
40ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Sound quality – Good
Video quality – Fair
Talking head A/V sync – Poor
Probably not usable
Sound quality – Fair
Video quality – Fair
Talking head A/V sync – Poor.
Probably not usable
80ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Sound quality – Fair
Video quality – Fair
Talking head A/V sync – Poor.
Probably not usable
Sound quality – Fair
Video quality – Fair
Talking head A/V sync – Poor.
Probably not usable
150ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Not tested.
Not tested.
200ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Not tested.
Not tested.
500ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Not tested.
Not tested.
1000ms
Sound quality – Good
Video quality – Fair
Talking head A/V sync – Generally drifts out of sync.
Probably usable
Not tested.
Not tested.
2000ms
Not usable
Not tested.
Not tested.
Bandwidth set: 1.5 Mbps Audio & Video Results
Total Latency (RTT)
Packet Loss Set
0%
0.5%
1%
20ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Sound quality – Good
Video quality – Fair
Talking head A/V sync – Poor
Probably not usable
Sound quality – Fair
Video quality – Fair
Talking head A/V sync – Poor.
Probably not usable
40ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Sound quality – Good
Video quality – Fair
Talking head A/V sync – Poor
Probably not usable
Sound quality – Fair
Video quality – Fair
Talking head A/V sync – Poor.
Probably not usable
80ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Sound quality – Fair
Video quality – Fair
Talking head A/V sync – Poor.
Probably not usable
Not tested.
150ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Not tested.
Not tested.
200ms
Sound quality – Good
Video quality – Good
Talking head A/V sync – Fair
Not tested.
Not tested.
500ms
Sound quality – Good
Video quality – Good
Desktop – significant lag
Talking head A/V sync – Fair
Not tested.
Not tested.
1000ms
Sound quality – Fair
Video quality – Fair
Desktop – significant lag
Talking head A/V sync – Generally drifts out of sync.
Probably usable
Not tested.
Not tested.
2000ms
Sound quality – Poor
Video quality – Fair
Desktop – unresponsive
Talking head A/V sync – Generally out of sync.
Not usable
Not tested.
Not tested.
Lync Tests Run in a LAN Environment with UDP Audio Selected
- Lync audio conferencing quality remained fair and was usable down to 5mbps/2000ms RTT/1% packet loss. However, with these RTT and packet loss settings it was difficult to interact with the desktop regardless of audio usage.
- Lync audio and video conferencing remained fair and was usable down to 5mbps/80ms RTT/1% packet loss.
Lync Tests Run in a WAN Environment where UDP Audio was not available.
(WAN environment includes an Access Gateway running is Secure Gateway mode)
- Lync audio conferencing quality remained fair and was usable down to 768kbps/200ms RTT/1% packet loss.
- Lync audio and video conferencing remained fair and was usable down to 1.5mbps/500ms RTT/0% packet loss.
This document applies to: