Cell Relay Archive

[murray@pa.dec.com (Hal Murray)]

> > That's one way of looking at it.  Another way of looking at it
> > is that the single bit correction performs no useful function
> > because it doesn't significantly affect the packet loss rate
> > seen by the application.  (The story may be different for
> > non-packet data, admittedly.)
> > 
> 
> why do you say that?

There are 53 bytes in each cell.  5 bytes are header.  If we assume
the errors don't know about cells or headers, then 10% of the errors
will be in headers and can get corrected.

I'd call 10% not very significant.

It would get much more interesting if there was another error correcting
mechanism working on the payload.

Another way to look at this whole mess is that header correction is only
interesting if you are operating with an error rate that is high enough
so that corrections are helpful but low enough so that the link is still
useful.

-- 
These are my opinions, not necessarily my employers.