| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Dear Colleagues, We are discussing relative cost comparisons between existing 100G SR10 and proposed SR4 PMD in light of economic feasibility (EF) in this meeting, as evidenced in multiple contributions. I would like to raise some questions I have in the regard. Although IEEE does not specify, it seems to me that the eventual form factor is part and parcel of broad market potential (BMP) and indirectly economic feasibility (EF) of a PMD. From that point of view, it seems to me – and I seek correction – that SR10 is fundamentally impaired in its BMP relative to the proposed SR4, based on being instantiated in CXP vs. QSFP, respectively. Part of the success of the 40GBASE-SR4 will be due, at least in part, to the fact that SR4, LR4 and copper are all housed in the same QSFP form factor. A switch can support three media with one port type. Even though CFP (and CFP2 in time) is overkill for SR10 compared to 802.3ba specs, it is nonetheless being produced by start-ups and introduced by main-line transceiver vendors because there are CFP (and CFP2) ports available for 100GBASE-LR4.
Even telecom switch makers are qualifying SR10 in CFP because CFP is there for SMF and their customers are asking to support SR10 as well. BMP of a PMD cannot be divorced from how common a port-type is and how many PMDs use same form factor. It seems to a primary consideration. CXP on the other hand is primarily an electrical connection (including AOC) and is typically found on the back of chasses for interconnection to other chasses. It is not likely to be found on the front of a switch. It seems to me that
relatively high volume applications in the Ethernet world are limited for CXP, regardless of its predominance in Infiniband. (Is this a misunderstanding on my part? Why?) I do not see that it matters to EF whether SR4 in QSFP is 1.3x or 0.7x SR10 in CXP. Please correct my thinking. Warm regards, Robert |