Why is the Ethernet data frame size limited to 1500 bytes?

Question

I am still not sure why the Ethernet data frame size / MTU was set maximum to 1500 bytes. I found out a document from Xerox which is from 1980. Available at (http://www.pennington.net/archives/ethernet/Ethernet_Version_1.pdf).

At point 6.2.3 it says that

The data field contains a sequence of n octets, where 46 < n < 1500 Within this range, full data transparency is provided, in the sense that any arbitrary sequence of octet values may appear in the data field.

I am not fully understand the reason that provided in this documentation. How does full data transparency provided? Is there any calculations behind this?

Anyone can share to me?

Answer 1

"Data transparency" in this context means that Ethernet doesn't care what kind of data it transports. You put any payload in a frame, send it to the destination and extract the exact same payload.

Initially, the maximum frame size was a trade-off between efficiency and latency. The larger a frame, the lower the overhead, the better the effiency. However, early Ethernet was a shared medium, so any frame in transit occupied the whole segment - the longer the frame, the longer the network was blocked and other senders had to wait. Additionally, early Ethernet NICs required fast local memory for buffering a frame. Ethernet was aiming to reduce prices, so then-expensive RAM had to be kept at a reasonable minimum.

When switches were introduced in the early 1990s, the maximum size had to be kept for compatibility reasons: a sender has no way to tell whether the frames it sends are going over a shared medium (anywhere) or whether they are switched at all times. Hardware buffers have to sized to maximum frame size as well and can't change on the fly.