on the suitability of ping to measure latency · we use ping • but, what is the actual...
TRANSCRIPT
![Page 1: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/1.jpg)
On the Suitability of ping to Measure
Latency Cristel Pelsser <[email protected]>
Luca Cittadini <[email protected]> Stefano Vissicchio <[email protected]>
Randy Bush <[email protected]>
2013.05.13 Tokyo Ping 1
![Page 2: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/2.jpg)
We Use Ping • But, what is the actual distribution of
the RTTs? • Could there be a rich distribution? • We were using Atlas and found ping
variance to be unexpectedly high • So we decided to calibrate our tools • So we ran a paris traceroute series from
Roma Tre to Ashburn 2013.05.13 Tokyo Ping 2
![Page 3: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/3.jpg)
Paris Traceroute Uses Flow-ID
To Explore Hashed ECMP / LAGged Alternative Paths
2013.05.13 Tokyo Ping 3
![Page 4: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/4.jpg)
2013.05.13 Tokyo Ping 4
![Page 5: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/5.jpg)
Roma Tre - Ashburn
2013.05.13 Tokyo Ping 5
Notice Linux Rounding
![Page 6: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/6.jpg)
ECMP != Equal Latency Multi-Path
2013.05.13 Tokyo Ping 6
![Page 7: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/7.jpg)
So, A Simple Path
2013.05.13 Tokyo Ping 7
Dallas Ashburn
Single Layer Three Path
![Page 8: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/8.jpg)
2013.05.13 Tokyo Ping 8
Simple Path Used LAG
FreeBSD Not Rounding
![Page 9: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/9.jpg)
Looking for Causes • Very Reproducible • Circuit Loading <= 50% • Same different times of day • Same different probe timing/spacing • Same UDP or ICMP
• Layer < 3 Hashing => Large Effect
2013.05.13 Tokyo Ping 9
![Page 10: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/10.jpg)
Layer-3 Only Hash
2013.05.13 Tokyo Ping 10
![Page 11: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/11.jpg)
Event During Run
2013.05.13 Tokyo Ping 11
![Page 12: On the Suitability of ping to Measure Latency · We Use Ping • But, what is the actual distribution of the RTTs? • Could there be a rich distribution? • We were using Atlas](https://reader033.vdocuments.site/reader033/viewer/2022042711/5f77fffcf1163d15b0711187/html5/thumbnails/12.jpg)
Selected Summary
2013.05.13 Tokyo Ping 12