Denver, All,
I'd like to both (1) comment on DSLAM classifications and DSL nomenclature in general, and (2) ask a general question of yourself and the board concerning this matter.
Your depiction of the DSLAM falls into one of two general classifications, from a protocol perspective, according to some industry word smiths, trade journalists and practitioners. A while back, some of these players attempted to delineate between the DSL Access "Multiplexor" (DSLAM), and the DSL "concentrator."
The variant that you described falls into the concentrator class, if we are to put any credence into this classification scheme, since it uses frames and routing in a contention scheme of the "best-effort" type on the back-end (toward the network's core, whatever that network happens to be), instead of a more "deterministic" form of scheme which uses ATM switching.
In brief, they contended that a Multiplexor = Switching, and a
Concentrator = Routing.
Explanation and derivation: The classical DSLAM model, as it was defined by Bellcore and numerous other ITU participants including Alcatel, NT, LU (AT&T at the time), etc., prior to the time that Internet influences were even conceived, defined a backplane that collected and managed 53-byte cells at the convergence sublayers, not Ethernet frames, and handed these cells off to an ATM fabric in the network's edge, towards the core. With time, the influences of the Internet Protocol became felt, until the implications of IP were profound, and eventually manufacturers both saw the need for, and in some cases were pressured into, modifying their designs to incorporate through substitution, a contention based scheme using Layer 2 frames on the backplane, instead of deterministic ANSI-defined cells.
And in turn, at the gateway they incorporated packet routing in place of cell switching.
In the concentrator mode, these frames, in turn, were collected and distributed in a manner that you described in your post as follows:
>>It [DSLAM] aggregates the Ethernet traffic from the modems and concentrates it into a single Ethernet backbone to go into a router, gateway or switch.<<
The question that I'd like to ask is this: Has this distinction taken hold in the industry? Or are both forms (cell-switched- muxes and frame-contention- concentrators) considered to be DSLAMs at this point in time, without distinction?
Regards, Frank C. |
