Irony on the rocks with a twist...

Once upon a time, the three major branches of the U.S. Military placed a lowest bidder contract for computers...because these branches had their own procurement offices and funding, the Army ends up with DEC computers, the Air Force ends up with IBM computers, and the Navy ends up with Unisys computers. It would not be until the U.S. invaded Grenada that it was discovered that these three different systems were proprietary, and thus incompatible with one another, meaning they could not communicate or share data (ah, yes...that low-bid government bureaucratic redundancy at its finest). Since no two branches could agree on scrapping their systems in favor of the other one, another solution had to be found. This is where TCP/IP comes in.

TCP/IP is a suite of network protocols devised to allow different kinds of information systems to share data over a network. This required it to be system-neutral in order to allow it to cross the different proprietary platforms. It solved the problem and became an instant success with the DOD, and thus was available for integration into then fledgling Internet. At the time, nobody could have predicted that we would be where we are now. The explosive growth of the personal computer, the Internet, broadband, commercial networks, the World Wide Web, wireless networking, and commercial VoIP, have all served to “stretch” TCP/IP to cover situations for which it just wasn’t designed for. The reason TCP/IP is still around is due to the fact that it is fairly simple and, despite its limitations, it still does work, however, it does have many limitations and potential vulnerabilities that are not addressed through its current configuration.

Since TCP/IP is really a marriage of protocols, it is important to understand what each part does and why:

TCP - is responsible for verifying the correct delivery of data from client to server. Data can be lost in the intermediate network. TCP adds support to detect errors or lost data and to trigger retransmission until the data is correctly and completely received.

IP - is responsible for moving packet of data from node to node. IP forwards each packet based on a four byte destination address (the IP number). The Internet authorities assign ranges of numbers to different organizations. The organizations assign groups of their numbers to departments. IP operates on gateway machines that move data from department to organization to region and then around the world. http://www.yale.edu/pclt/COMM/TCPIP.HTM

The implications of making such a change to the protocols which comprise the "foundation" of modern computer networking is so great that updating these two protocol suites are being accomplished separately. It takes time for a global consensus to reach an agreeable standard concerning protocols.

Acceptance for replacing IPv4 with IPv6 is well on its way, however, no such convergence has taken place concerning updating TCP.

The problem of a new TCP is as complex (if not more so) that the TCPIPv4/v6 changeover which the Internet community has found very hard to deal with.

However, the factors in slow IPv6 deployment largely revolve around the fact that there is no communicated compelling reason to change. Given that a point of time will arise when changes to TCP are necessary for basic performance, it can be expected that, if a migration is conducted with appropriate change management planning, the adoption will be far quicker and far smoother than the IPv6 changeover. However, some basic factors need to be taken into account:

A compelling reason to change has to exist - No major change to core Internet protocols is going to be undertaken unless a compelling reason exists.

A compelling reason to change has to be communicated - Similarly, a major communication project is needed for a change of this magnitude. Trying to do it without formal change management planning would not succeed.

Involvement of major players will help - In particular, major players should be engaged, and planning for total adoption should take into account their adoption schedules. Once again compelling reasons to change have to be communicated: the cost to the market leaders to make the change across all product ranges is substantial, and the economic reasons to stay with current configurations may be compelling. We may be talking about changes to every connected device on the planet here. This could exceed 1 billion devices.

Co-existence with old systems has to be allowed but only for a limited period of time - There cannot be a one day changeover. This is addressed elsewhere in the Internet Analysis Report - 2004, but for a quite substantial period of time the new system must co-exist with the old. http://www.circleid.com/print/752_0_1_0/

Generally speaking, changes to technology are driven by reaching a "critical mass" of sorts. TCP has not yet reached this critical mass, but the continued development of IPv6 is bound to highlight TCP's shortcomings.

This should by no means imply that nobody is working on TCP's replacement:

Popular potential replacements for TCP
SCTP or Stream Control Transmission Protocol, initially developed by Randall R. Stewart, and Dr. Qiaobing Xie, this has gotten a very warm response from the international community.

DCCP, or Data Congestion Control Protocol, this grew out of UDP, or User Datagram Protocol development.

eXplicit Control Protocol, or XCP, an exciting protocol being developed at the Informational Sciences Institute located at USC

SACK-TCP, or Selective Acknowledgement Transmisssion Control Protocol, out of the Advanced Networking Division of The Pittsburgh Supercomputer Center, a joint effort between the University of Pittsburgh and Carnegie Mellon University.

Additional replacement protocols in various stages of development

S-TCP, or Scalable Transmission Protocol at Cambridge University

FAST TCP, Fast Transmission Control Protocol out of Caltech

BIC-TCP, or Binary Increase Congestion Transmission Control Protocol at the Computer Science Department of NC State University

HS-TCP, or High Speed Transmission Control Protocol at the Center for Internet Research, which is part of The International Computer Science Institute

HS-TCP-LP, or High Speed Transmission Control Protocol Low Priority, at Rice University.

Westwood + TCP an improved TCP Reno (see below) entry by a technical university in Italy called Politecnico di Bari

There are also various experimental TCP implementations named after various geographic locations, (but these seem more like experimental "proof of concept" types of implementations):

TCP Reno
TCP Tahoe
TCP Vegas
TCP NewReno
TCP Net Reno
TCP Niagara

Additional Semi-Related (and exciting) Resources

Sigtran.org- Signal Transport Protocol Website

Technical Data Center

The Information Technology Professional's Resource Center

High Speed Internet Protocol Test Results

Web 100 Project

Net 100 Project

TANSTAAFL!