The rapidly growing Internet puts more and more demands on the transport
protocols in use nowadays. The solutions for used for data transmissions were
developed near the beginning of the creation of the Internet, using appropriate
techniques at that time, so now they’ve reached the limits of their efficient operation
and in a variety of network environments they’re unable to provide acceptable
performance for the users.
Recently a good number of theories were proposed for improving and adapting
to the needs of the present the traditional TCP congestion control mechanisms, and
investigations are underway to create new protocols which can replace TCP altogether.
These new mechanisms and TCP versions have appeared in the latest versions of Linux
kernels and new versions of Windows, which has resulted in today's heterogeneous
network traffic control mechanisms working side by side (or "against each other"). The
study and extensive testing of their functioning and interaction are of course essential
for future application of these versions.
In this paper I intend to examine the workings of some of the more important
TCP versions which can be found in the latest Linux kernels in a high-speed network
environment, using different measurement parameters and topologies. I also intend to
assess the viability of a new approach to transport mechanisms where the reliable data
transfer is granted by error correction and redundancy. The main aim of my
investigations is the analysis and evaluation of their performance, to see whether the
new protocols are capable of replacing the traditional TCP.