Evaluation

Question 1

The primary goal of TFO is to

Answer 1

eliminate one RTT of extra latency, thereby improving the performance of short flows

Question 2

tool to benchmark the web page download latency

for TFO-enabled Chrome and for standard Chrome

Answer 2

Google web page replay tool

Question 3

popular configuration as found by

Netalyzer

Answer 3

In our experiment, we emulated a broadband user with
4Mbps downlink and 256Kbps uplink bandwidth and with
a 128KB buffer

Question 4

Since all connections use the loop- back interface,

Answer 4

the TFO-enabled browser always has a valid TFO cookie and thus sends the cold HTTP requests in the SYN packet.

Question 5

TFO usually improves the PLT when the RTT is

Answer 5

high

Question 6

gains from TFO are expected to be small when RTT

Answer 6

small

Question 7

The 200ms RTT fig- ures roughly correspond to the expected performance on

Answer 7

mo- bile devices since mobile RTTs are typically on the order of 100–200ms

Question 8

server CPU utilization is nearly the same with and without TFO

Answer 8

in fact the CPU utilization was marginally lower when using TFO between 2000 and 5000 requests per second. We attribute this to the fewer packets that the server has to process when TFO is enabled as the request is included in the SYN packet.

Question 9

with the use of TFO

Answer 9

Average rate of transactions will be more than regular TCP