Bandwidth Galore (Update)

In a story entitled “Fujitsu Tests Optical WDM Communication Spanning 7,400km at 2.4Tbps”, published by Nikkei Business Publications, Inc., we see some impressive but not unexpected news. In Net Attitude, I talked quite a bit in the chapter called Fast about how we would have “Bandwidth Galore”. I also wrote a 1998 reflection by that name that is here.

Fujitsu Laboratories Ltd. has already commercialized a 1 Tbps optical underwater cable system that multiplexes a 10-Gbps signal into 105 channels. In other words they are enabling more than one hundred separate “windows” of light to simultaneusly and independently operate in a single strand of optical fiber. The newer technology, which Fujitsu hopes to commercialize in 2005, utilizes 240 “windows”. With each “window” handling ten billion bits per second, the aggregate through the single fiber comes out to 2.4 trillion bits per second — 2.4 terabits per second.

Lets put 2.4 terabits per second in perspective. The book Net Attitude has 472,000 characters in it (including spaces). Using eight bits to represent each character that would be 3.8 million bits in the book. At a speed of 2.4 TB/S it would be possible to transmit 633,000 copies of the book in one second. Lets take a CD. A CD can hold approximately 600 million bytes. With eight bits per byte that would be 4.8 billion bytes. At 2.4 TB/S it would be possible to transmit 500 CD’s full of data in one second!

That is the gee whiz part. The important part is that we will really need all this bandwidth. Imagine a bank with trillions (terabytes) of bytes of customer and financial data in digital storage. This is today’s reality. It is growing rapidly. How does one keep backup copies of vast amounts of data so that all of us can be assured that it is protected under any circumstances? One way is by having multiple redundant storage sites in various physical places around the world and then using optical networks to continuously make backup copies of the data in these multiple places. No single point of vulnerability. If one location is destroyed, the others are intact. Perhaps never more than a few seconds of lost data. Perhaps no lost data exposure at some point as every transaction that occurs is instantly recorded at a backup site. The transaction is not validated until the backup copy is made.

IBM and others have been doing this kind of redundant recording inside of storage devices for years. With vast bandwidth available, such as what Fujitsu has announced in November 2001, it will possible to do the same thing with the storage devices in effect being split geographically onto different continents. The Fujitsu demonstration was made over a distance of almost five thousand miles. This is the kind of protection of data that will be required as we become more and more dependent on digital data. Some people believe there is a glut of bandwidth. Yes, we will have bandwidth galore, but I think we are going to need it.