This is a discussion on Effect of link-layer compression on TCP bandwidth within the Linux Networking forums, part of the Linux Forums category; Hi, I will really appreciate any help here: 1. What kind of tools can i use to measure TCP bandwidth ...
|
|||||||
| FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
|
|||
|
Hi,
I will really appreciate any help here: 1. What kind of tools can i use to measure TCP bandwidth I have been using iperf, but if people have better opinions, please let me know 2. I want to analyze what effect link-layer compression might have on TCP bandwidth. I think i can again use Iperf for this, since link layer compression only compresses the headers. Is this a correct assumption? 3. I want to analyze effect of link-layer compression schemes that compress the entire packet rather than just the header. What tool can i use for this.. or what would be the most representative streams that i can use to get valid results. Bear in mind that compression depends on input entropy. Hence if the packets that are created by the tool have low entropy, they will be compressed more resulting in smaller size and higher bandwidth, but if the packets have high entropy, then they will be compressed less, resulting in larger size and lesser bandwidth. Thanks in advance. Shashank |
|
|||
|
Don't forget to analyze latency. As long as (a) The device in question has the computational power to (de)compress the data as fast as it can receive/send it, and (b) the data isn't relatively random, bandwidth will only go up. However, implementing compression on high-bandwidth, low-latency lines is likely to have a relatively large hit in latency, which in many instances, can actually be *more* damaging than a hit in bandwidth. steve |