* Bandwidth efficiency took on new dimension with packet switching
As we mentioned last time, there was an entire industry built around using TDM bandwidth efficiently due to the compelling economics of building corporate private-line networks. By the late 1980s, however, packet switching technologies, ranging from “statistical multiplexers” to frame relay to ATM, began to take off.
In particular, these devices started taking advantage of the asynchronous nature of data traffic. By “asynchronous,” we mean that there are gaps between transmissions, and these gaps can be filled in by traffic from other data sessions. The industry was waiting for the electronics industry to develop sufficiently fast and cheap chips to make this level of additional processing economical.
By a different measure, this asynchronous transmission, generically called packet switching, marked a shift in emphasis to a different type of “bandwidth efficiency.” As we pointed out last time, it’s nearly impossible to beat the efficiency of a T-1 carrying PCM voice, assuming that all conversations are connected continuously. However, the efficiency of TDM transmission comes into question when there are gaps in the traffic. With TDM, the bandwidth is reserved whether it’s needed or not.
Thus, we have the beginning of controversy as to whether packet switching is more “efficient.” And exact calculations have to be given up in favor of suppositions. Where TDM transmission is better than 99% efficient for a constant data stream, packet streams have what must be referred to as “variable efficiency.” The lighter the traffic and the larger the packets, the more efficient they are.
For instance, frame relay has a minimum of five, eight-bit bytes (octets) of overhead per frame. So if you’re only transmitting five bytes of information as the payload, you have 100% overhead – or only 50% efficiency. As the packets get larger, the efficiency gets better. If you have 95 bytes to transmit in the payload, you’re up to 95% efficiency for that particular packet.
But the real “bandwidth efficiency” here comes from the ability to share bandwidth among multiple data sessions. Even though the per-packet bandwidth efficiency is reduced, the aggregate efficiency is increased.
But this timeframe also marked the beginning of the “battle of the protocols.” In particular, a major defining characteristic of frame relay, ATM, and IP is the efficiency with which these protocols transport information. More and more, these efficiency arguments are starting to fade into oblivion. We’ll tell you why next time.
