| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
A key to the 40GE 10km SMF PMD discussion is understanding suitability of proposed approaches to different applications.
The primary, high volume application is Data Center and Campus links, with viability of any 40GE PMD solution highly sensitive to cost and time to market. Appropriately the focus of the debate within 802.3ba has been on cost and development risk, with the recognition that any solution must fully address the needs of the Data Center and Campus environment otherwise the entire rationale for having this PMD objective goes away.
The statement of Broad Market Potential in support of adoption of the 40GE 10km SMF PMD objective gives details of this primary application. "Servers, high performance computing clusters, blade servers, storage area networks and network attached storage all currently make use of 1G and 10G Ethernet, with significant growth of 10G projected in ’07 and ’08. I/O bandwidth projections for server and computing applications, including server traffic aggregation, indicate that there will be a significant market potential for a 40 Gb/s Ethernet interface."
High volume deployment of 40GE interfaces creates opportunities for other applications, in particular OTN clients. There has been much less discussion of this in 802.3ba (again appropriately) resulting in past assumptions about different technical approaches left unexamined, specifically that Serial is better suited than WDM as an OTN client.
Reliability
40G Serial would appear to have 4x better reliability than 40G CWDM because it has one laser instead of four. It might further be argued that its reliability is additionally improved because it is cooled. The exact quantification of the ratio of reliability of four 10G DFBs to one 40G EML is a complex problem. However, both have such negligibly low absolute failure rates that a single integer reliability ratio makes this an unimportant consideration.
Neither 40G Serial nor 40G CWDM optics reliability is an issue.
Operation and Maintenance
A perceived advantage of 40G Serial approach is that it lends itself to simple installation and testing methods. In particular link viability can be checked with a power measurement. This is in contrast to perceived drawback of the CWDM approach which appears to need complicated per wavelength measurements to detect link problems. There is some validity to this concern, however a majority of problems can be identified with a simple power measurement for either CWDM links or Serial links. This is because it is not individual laser or channel failures inside the optical module that cause majority of problems but rather physical interconnect failures like broken fiber, dirty connectors, failed power supplies, or improper configuration. Because of the high reliability of 10G DFBs that will be used for CWDM, there is no need for routine per wavelength field measurements. For the rare cases where diagnotic measurements are required, a CWDM link that can be measured with low cost and common 10G test equipment with the addition of a low cost CWDM. This is in contrast to diagnostic measurements for 40G serial which require very costly test equipment.
Majority of operation and maintenance problems affect 40G Serial and 40G CWDM in the same way, and conventional power measurement methods can be used for both.
Alarms and Monitoring
There is a perceived advantage that 40G Serial alarms and monitoring is simpler since it is only done for one wavelength, versus having to alarm and monitor four wavelengths. This in fact is not the case. Even though the CWDM approach uses four wavelengths, the optical module is treated as a single entity and all failures or problems are treated as affecting the entire link, no different than for Serial.
For the system, alarms and monitoring of 40G Serial or 40G CWDM interfaces are the same.
Electrical Interface Protocol
Both 40G Serial and 40G CWDM have a 4x10G electrical interface. Neither is presently usable as an OTN client. To remedy this situation, multiple companies (ALU, CISCO, Cortina, Finisar, Nortel, NTT) made proposals to the ITU-T to develop a mapping to MLD for OTN applications to deal with skew associated with multiple electrical and potentially multiple optical lanes. These proposals have been well received, and are under active discussion in the group responsible for OTN protocols (SG15/Q11.)
OTN MLD is equally applicable and necessary to use 40G Serial or 40G CWDM PMD as OTN client; neither offers an advantage here.
Extension to OTN line side (DWDM)
A desire has also been expressed to extend 40G Ethernet technologies not just to OTN clients but also to OTN line side interfaces, with Serial appearing to offer an advantage. However, the proposed 40G Serial PMD approach is EA modulator/DFB laser based, which is not usable for greater then 2km reach in C or L band because of dispersion effects. At 1550nm, 40G Serial EML technology can not even be used for 10km reach, and certainly can not be used for 40km or 80km C or L band metro reaches.
Both 40G Serial and 40G CWDM Ethernet technologies are equally not extendable to OTN DWDM applications.
Conclusion for OTN applications
The most important consideration for leveraging 40GE 10km SMF PMD for OTN is that 802.3ba adopts a technology which will be highly successful in its primary Data Center and Campus link application. Any burden that adds cost or delays deployment of the primary application, will have significant negative impact on OTN use, and will reduce the value of using a high volume Ethernet product for other applications.
I would like to encourage those interested in leveraging 40GE 10km SMF PMD technology for OTN applications to confirm or correct the above considerations, so that we can fully understand their perspective.
Chris
From: John DAmbrosia [mailto:jdambrosia@xxxxxxxxxxxxxxx]
Dear Task Force Members, Per Motion #9 from July, the editorial team is working on creating a “a draft based on adopted baseline proposals for circulation prior to the September 2008 interim meeting .” Unfortunately, at the July meeting the Task Force did not reach consensus on a baseline proposal to satisfy the 40G over => 10km SMF objective. Therefore, in September we need to reach closure on this issue.
With that said, I would like to strongly recommend that the TF make use of the reflector to discuss the various issues of debate that have been going on, both during the meetings and during offline discussions.
Let’s use the next several weeks to have meaningful debate so we can reach consensus at the September meeting.
Regards,
John D’Ambrosia Chair, IEEE P802.3ba Task Force |