How to calculate the throughput in TCP? - tcp

I am trying to calculate the throughput for TCP/MPTCP by using the parameter tp->packets_out and also tp->snd_una but they are not accurate.
How does wireshark do it?
Or does anyone know any solution?
Thanks in advance.

I am the author of https://github.com/teto/mptcpanalyzer which tries to provide some statistics for MPTCP connections. Current release has many bugs but the upcoming one is much better. Feel free to open issues there if you have any problem.

Related

Neighborhood collectives-reduce operation

I need the allreduce (MPI_Ineighbor_allreduce) for collective communication in MPI, unfortunately it is not included yet.
The obvious, not so efficient solution, is to use MPI_Neighbor_alltoall
at the expense of increasing buffer size, Do you have any suggestions?
Is there any plan to include this in future releases?
Thanks
there is no such thing as MPI_Neighbor_allreduce in the MPI standard.
if you want it, feel free to ask for it at http://mpi-forum.org/
btw, did you mean to use MPI_Neighbor_allgather instead of MPI_Neighbor_alltoall in order to implement the neighbor allreduce ?

Does NS-3 supports Mobile IPv6?

I have to simulate a few scenarios of network mobility (MIPv6). I was told to use NS-3, but I don't find any good information about it. Can anyone give me some lights about this? I just need to simulate a few simple scenarios of networks moving around...
Thanks anyway.
As far as I can tell, there are two ways to do ns-3 MIPv6 simulations:
using DCE: http://www.nsnam.org/~thehajime/ns-3-dce-umip/getting-started.html
using a native UMIP implementation from http://eudl.eu/pdf/10.4108/ICST.SIMUTOOLS2010.8682 but I do not believe that it has been released.

Is it possible to port recent SQLite to VxWorks 6.8?

I am struggling while trying to port recent SQLite sources to VxWorks 6.8. The architecture is PPC.
I made a separate topic (Crash around pthreads while integrating SQLite into RTP application on VxWorks) to provide all details about the particular problem I am experiencing at the moment. But it looks like the problem is too specific and requires some amount of certain experience (porting C code to different platforms, pthreads, SQLite, knowledge of VxWorks).
So, I decided to just get a confirmation that it is doable at all. I mean it is doable for sure but I need to know that someone has actually succeeded with it within some reasonable time frame.
Please respond only if you yourself accomplished this. No general suggestions, like: "VxWorks is POSIX, SQLite is C - should not be a problem".
To moderators: I don't mean to duplicate my question. I am just norrowing it down and intend to close if no constructive ansver(s) appear.
Thanks in advance
Ok, I have figured it out - it was the problem in dosFS. I formatted the flash to HRFS and was able to run SQLite.
I had few porting issues on the way, something that is really platform-dependent. One familiar with POSIX functions should be able to figure it out.
I guess there is a problem with dosFS. So for now I will stick to HRFS.
In case of any questions about the porting itself - just contact me, may be I had te same and fixed.
Just for informaiton, I am using PPC and VxWorks 6.8.
Regards

creating a loss system for voice data in Linux

how do I create a system that takes in voice data(rtp) and then creates loss in this data(like delay or packet drop/loss)? The output of the system(data) should be readable which made me think i might not be able to use ns-2.Also, ns-2 does not support VBR(needed for voice). I might be wrong in this aspect though. How can I achieve this loss condition in linux environment? please give suggestions.
NIST Net looks as if it will do what you want.

Easy way to measure LAN network speed of a partcular computer?

I want to measure the network speed between two PCs on Local Area Network. (I'm thinking of getting a Network Area Storage (NAS) device and I want to see how fast the current setup is to get an idea of how fast the NAS needs to be.
I'm thinking I'll just copy some files and look at how long that takes, but I thought there might be a more precise way to measure that.
To measure the network performance, iperf is the best tool I have seen.
I do not believe this will be helpful in solving your problem, however. You aren't going to be able to size a NAS solution based on the network performance.
Transferring a file using the same protocol the NAS would is certainly more relevant for this case. Still, not helpful.
You can assume that any modern machine can transfer data at over 90% of the link speed. (whether involving the local storage or not, unless there's a problem or significant other load).
A dozen reasonably modern clients should be able to crush a fairly powerful NAS. You really need to simulate real-world load. That's basically imposssible.
You need to approach sizing your NAS solution from a different perspective. The capabilities of the clients really aren't relevant.
This might work if you do the same process with another computer. It might not be precise, but could give an idea.
Each NAS device tends to produce as part of its spec a throughput number. Generally speaking if speed is of concern, you would be better off which choosing SAN over NAS.

Resources