Virtual Transport Layer, a protocol deployment management system aiming (1) to dynamically deploy protocol mechanisms/components, in order (2) to ensure optimal data moving between end-points.
I am spending time bettering the implementation. Until now the focus has been on the correctness rather than performance. I will keep correctness, but I will improve performance, especially in light of practical experience with applications. Suggestions are welcome.
On Ubuntu 20.04 do
sudo apt install libelf-dev pcap-dev libndpi-dev gcc-multilib
mkdir bin
make
> Hooker contains two programs:
- The daemon in /hooker folder, that is created with a make call. It is user prog part of Hooker.
- The kernel prog part that contains mainly two eBPF progs: sock_ops, to monitor specific or a set of TCP applications, and sk_msg, to redirect the monitored applications' data.
> First: build and launch (deploy) kernel prog part:
$ make
$ cd bin
$ sudo ./vtl_ui
Follow the instructions and make choices.
> Second: start user daemon
$ cd hooker
$ make
$ sudo ./hk-daemon <IPPROTO> <HOST_MODE>
IPPROTO: this argument is an integer and is used to indicate which protocol should be used to replace TCP. Indeed, VTL allows the user to manually fix the substitution protocol by specifying one of the supported IP protocol (list below) or let this task to VTL by specified 253 (default) as the argument value.
List of substitution protocol supported by VTL and their corresponding IP number:
Protocol | IP number [IANA] |
---|---|
UDP | 17 |
UDPLite | 136 |
DCCP | 33 |
SCTP | 132 |
QUIC | 142 (conventional in VTL) |
VTL | 253 (conventional in VTL) |
HOST_MODE: a string to indicate if the hooker daemon will hook either a TCP client or a TCP server.
> Finally, just run your TCP applications.