A Minor NIT
The SPE content is considered bytes when being passed through
the SONET network. When the SONET network needs to "adjust"
1 SPE with respect to another SPE in order to account for
clock jitter and tolerances, it does so on byte boundaries.
This could potentially further separate a 66-bit word. The
SONET network also uses byte boundaries to find the OH bytes
within a frame.
While I generally agree with the above statement I would
like to point out one possible NIT regarding SONET pointer adjustments.
I prefer to view, and please feel free to jump in if you disagree, the SONET
STS-N Frame structure, and especially its encapsulated SPE structure
as a continuous stream. Now pointer adjustments do cause short term
instantaneous variances in the data rate, but in the end the average rate
will be that of the associated Stratum Clock reference.
A positive pointer adjustment will add one stuff byte, and thus
one payload byte from this SPE. A negative pointer adjustment will
delete one stuff byte, and thus add one payload byte to this SPE,
actually the H3. Please note that in either case all of the payload bytes
get there, eventually.
But in the end the SPE stream is concatenated together logically
to form a continuous stream of bytes. The parsing of 66b encoded frames
will occur on this stream of bytes and thus should be un affected by
activities at the SONET physical layer including pointer adjustments
and frame structure.