Are you tired of slow networks? Well Fujitsu thinks that they have the answer!
Many network protocols run over Transmission Control Protocol (TCP), basically a set of rules for network communication. TCP is an essential part of the greater Internet Protocol (IP), which is more of the same. Thus you’ll often see them referred together as TCP/IP, as they’re almost always joined at the hip. This is basically the heart of what makes Ye Olde Internet work, and what allows programs running on computers (and other gadgets and gizmos) to communicate with one another and the world at large.
What TCP/IP does is break down what you want to communicate (be it an email, the contents of a file, or what have you) into tiny chunks called packets. It then sends these packets through the many varied paths through a network from point A (you) to point B (the recipient). If you were to send the whole shebang in one chunk, chances are that big piece would get damaged or stuck along the way, and then you’d have to resend it. Quite possibly again and again and again. Which would be slow and unwieldy. Because in today’s world networks are thousands upon thousands of computers shaking hands, routing information down this way or that depending on which looks to be less utilized, et cetera ad nauseam. Back in the good old days when the internet was shiny and new, this wasn’t such a challenge. Today though it’s like tightrope walking across a congested superhighway of tightropes over a minefield while carrying a cup of water and hoping to not spill a single drop along the way. By breaking your message into lots of tiny packets carried by multiple tightrope walkers and reassembling them at the end it allows to send packets down multiple paths and to request only to resend the damaged or lost ones, making communication from one end of the Earth to the other possible amidst the chaos.
The thing is, you see, TCP/IP is basically as old as the internet itself, more or less. And while it works, there’s always room for improvement.
An muck less often used alternative to TCP is the User Datagram Protocol (UDP), which was tailored specifically for streaming media, such as a movie. It’s designed for minimal delays and getting things from point A to point B a lot faster, but a lot less safely. The constant handshaking and quality checking of TCP is virtually non-existent in UDP, by design. Instead UDP just machineguns data as rapidly as possible. It makes it a lot faster, but a lot more dangerous for really important things.
That’s why most things still use TCP/IP. It’s safe. It’s reliable.
But it’s also slow.
That’s where Fujitsu comes into our picture. They reckon that they have a solution that’s the bee’s knees to TCP/IP. Because of the minimal overhead involved in UDP they built their new network protocol around that. But instead of using unprotected UDP directly, they made themselves a man in the middle. To computers it looks like a TCP/IP connection. To the networks that shuffle the data around it uses a UDP port. But thanks to their man in the middle, it’s kind of both, basically.
The advantage that Fujitsu brings is that they’re intelligently hybridizing TCP with UDP and adding in some of their own special sauce to fix flaws in both in the process. For example, UDP often sends too much data through, congesting the bandwith and clogging up attempts to communicate. Whereas Fujitsu’s approach tries to intelligently limit how much data to throw down the pipe at one time, allowing more information to pass through by not hogging all of the bandwidth. And more importantly, Fujitsu has designed a way to tell the difference between packets that have been lost, and packets that have been dropped, an important distinction that can prevent unnecessary requests for retransmitting a packet. The less duplicate and useless retransmissions of data is the less bandwidth wasted in the process of getting everything from point A to point B.
All-in-all, data transmission tests Fujitsu has run from Japan to the United States using their new technology have seen thirty times more speed and a latency just one-sixth of the time you get with TCP/IP. That’s pretty impressive, and if everyone started using that, we’d see a whole new internet.
To that end, they hope this will make their new network protocol become a key player in today’s data-choked world now that our mini-devices do everything. Look to see Fujitsu start commercializing their product sometime this year.