Sachin Agarwal‎ > ‎Publications‎ > ‎

GIS 2011, Infocom 2011

Toward a Push-Scalable Global Internet

Push message delivery, where a client maintains an “always-on” connection with a server in order to be notified of a (
asynchronous) message arrival in real-time, is increasingly being 
used in Internet services. The key message in this paper is that 
push message delivery on the World Wide Web is not scalable 
for servers, intermediate network elements, and battery-operated 
mobile device clients. We present a measurement analysis of a 
commercially deployed WWW push email service on  Android phones to highlight 
some of these issues. Next, we suggest content-based optimization 
to reduce the always-on connection requirement of push m
essaging. Our idea is based on exploiting the periodic nature 
of human-to-human messaging. We show how machine learning 
can accurately model the times of a day or week when messages 
are least likely to arrive; and turn off always-on connections 
these times. We apply our approach to a real email data set and 
our experiments demonstrate that the number of hours of active 
always-on connections can be cut by half while still achieving 
real-time message delivery for up to 90% of all messages.

This article was discussed in an MIT Technology Review blog by Christopher Mims.

Sachin Agarwal,
Apr 25, 2011, 10:57 AM
Sachin Agarwal,
Apr 25, 2011, 10:57 AM