作者: Ricardo Morla
DOI:
关键词:
摘要: HTTP response size is a well-known side channel attack. With the deployment of HTTP/2.0, estimation attacks are generally dismissed with argument that pipelining and multiplexing prevent eavesdroppers from finding out sizes. Yet impact actually have in estimating sizes has not been adequately investigated. In this paper we set to help understand effect web objects on Internet. We conduct an experiment collects TLS record 10k popular sites. gather evidence discuss reasons for limited amount used Internet today: only 29% HTTP2 observe pipelined 5% multiplexed. also provide worst case results under different attack assumptions show how effective simple model can be. Our conclusion especially yield, as expected, perceivable increase relative object error yet extent observed today simplicity current mechanisms hinder their ability estimation.