You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
However, there is no option to specify a "loss" factor that simulates packets randomly dropping.
The biggest meat/potatoes of QUIC that gives it the advantage over TCP is the fact that there is no head-of-line blocking when sending multiple streams in HTTP/2. In HTTP/2, 1 TCP connection is made and that flattens all parallelizable streams (video file (stream 1) + HTML page (stream 2)) into 1 multiplexed stream. In a lossy network, a big piece of media like a video getting sent over would block all other independent streams (like the HTML file). With HTTP/3 we don't have this problem.
We should showcase this advantage of QUIC vs TCP in our secnetperf tool.
Proposed solution
Add a "loss" option in the tool that allows us to see what the numbers for various scenarios look like when the network is not perfect.
Additional context
No response
The text was updated successfully, but these errors were encountered:
I think we need to be careful not to try to make one tool to rule them all. There are several ways out there to emulate loss on a network or network interface, and they can have various impacts on performance. So, I think it would actually be better not to have this built into secnetperf.
Describe the feature you'd like supported
Today, we use the secnetperf tool (https://microsoft.github.io/msquic/msquicdocs/src/perf/readme.html?q=secnetperf) to measure performance of QUIC vs. TCP on various platforms.
However, there is no option to specify a "loss" factor that simulates packets randomly dropping.
The biggest meat/potatoes of QUIC that gives it the advantage over TCP is the fact that there is no head-of-line blocking when sending multiple streams in HTTP/2. In HTTP/2, 1 TCP connection is made and that flattens all parallelizable streams (video file (stream 1) + HTML page (stream 2)) into 1 multiplexed stream. In a lossy network, a big piece of media like a video getting sent over would block all other independent streams (like the HTML file). With HTTP/3 we don't have this problem.
We should showcase this advantage of QUIC vs TCP in our secnetperf tool.
Proposed solution
Add a "loss" option in the tool that allows us to see what the numbers for various scenarios look like when the network is not perfect.
Additional context
No response
The text was updated successfully, but these errors were encountered: