Data encapsulation of HTTP, TLS, TCP and IP

数据包封装过程

 原文：

TLS record size has an important impact on the performance of HTTPS. Here we discuss the effect of record size on data encapsulation, as shown in Figure 2. TCP is a byte-stream protocol, which can split message from the upper layer of TCP (e.g., TLS) in arbitrary ways for transmission. Thus, there is a length field in TLS header to make the receiver know where the record ends. For large record size, TLS record exceeding MSS limit will be transmitted in multiple TCP payloads, and only one of these TCP payloads contains the TLS header.

While for small record size, the entire TLS record can be accommodated in a single TCP payload. Since the length of TLS record is usually less than MSS, TCP will intercept the next TLS record to pad the current payload as large as MSS.

As a result, a TCP payload may contain none, one or multiple TLS headers. On the whole, no matter which record size it is, we can always construct the relationship between packets and their corresponding HTTP messages.

译文：

TLS record size has an important impact on the performance of HTTPS. Here we discuss the effect of record size on data encapsulation, as shown in Figure 2. TCP is a byte-stream protocol, which can split message from the upper layer of TCP (e.g., TLS) in arbitrary ways for transmission. Thus, there is a length field in TLS header to make the receiver know where the record ends. For large record size, TLS record exceeding MSS limit will be transmitted in multiple TCP payloads, and only one of these TCP payloads contains the TLS header.

While for small record size, the entire TLS record can be accommodated in a single TCP payload. Since the length of TLS record is usually less than MSS, TCP will intercept the next TLS record to pad the current payload as large as MSS.

As a result, a TCP payload may contain none, one or multiple TLS headers. On the whole, no matter which record size it is, we can always construct the relationship between packets and their corresponding HTTP messages.

 

摘自论文：

Efficient Fine-Grained Website Fingerprinting via Encrypted Traffic Analysis with Deep Learning