One group, led by Agere Systems (NYSE: AGR) and Agilent Technologies Inc. (NYSE: A), is proposing an MSA called X2 (see X2 MSA Launched). The other group, led by Intel Corp. (Nasdaq: INTC) and Infineon Technologies AG (NYSE/Frankfurt: IFX), is proposing an alternative called XPAK (see Trio Announce 10-Gig MSA).
Things came to a bit of a head recently, when Intel announced its first XPAK chips and claimed that its move would result in “peace in our time,” as Neville Chamberlain once said (see Can Intel Make Transceiver Peace?). History seems to be repeating itself: War has broken out...
First off, Agilent mobilized its troops. In other words, it put together a statement on behalf of the vendors supporting the X2 MSA and sent it to Light Reading, in the hopes of stirring up a debate on the pros and cons of the two MSAs. Here it is, in all its glory:
10 Gigabit Ethernet Transceivers
Although currently the XENPAK MSA provides the standard for 10 GBE transceivers, there is much discussion about future smaller form factors and specs for these products. Customers have limited development efforts due to current market conditions, but indications are that industry demand for smaller, next generation designs may begin in earnest in early to mid 2003.
The “X2” MSA, announced earlier this year, and supported by the largest number of vendors (Agilent, Agere, JDS Uniphase, Mitsubishi, NEC, Opnext, Optillion and Tyco) in the industry, has the following advantages: 1. PCI compliance
X2 is designed to be PCI compliant in order to ensure the widest market applicability. PCI compliance is important and the MSA group has engaged multiple customers and their feedback indicates that they are confident of the X2 MSA PCI solution.
Pictures of the PCI solution are given on page 5 of the Introductory presentation on the X2 web site: http://www.x2msa.org/X2MSA_Rev0.9d4.pdf.
Additional technical design information detailing the PCI solution is given Section 6.5. and Section 6.2 of the X2 MSA at: http://www.x2msa.org/X2MSA_Rev0.9d4.pdf.
2. X2 has the broadest industry support
The X2 MSA was based upon the requirements and needs of a number of major players (top tier 10 Gb vendors) across multiple market segments.
The X2 membership includes most of the companies shipping XENPAK modules. These are the companies that defined & contributed to the original development of front panel pluggable 10 Gig optics.
Lightreading’s Webinar survey (archived here), indicated a preference of X2 to XPAK by a three to one margin.
3. Although much smaller, X2 leverages XENPAK
X2 uses the same electrical interface as does XENPAK. It also has a very similar approach for EMI containment. EMI design is usually an iterative process, and building (or qualifying) on a proven design saves time and resources for both the component vendor and end customer.
X2 has been designed to allow vendors who have developed XENPAK technology to directly apply this technology on X2. This will enable: - Faster time to market - Shared volumes lowers the costs of both platforms - Lower risk - More vendors to participate (more sources)
4. How does X2 leverage XENPAK?
The X2, as a derivative to XENPAK, uses some of the same component, electrical and especially EMI designs, which should make qualification easier, quicker and less costly for customers who have designed and qualified XENPAK.
In addition, the X2 MSA group has significant experience in higher performance modules used in telecommunications, which has been helpful in specifying a design that is robust in terms of thermal and EMI performance. This is important when using the design with longer wavelength lasers for longer distances or in applications where systems require 8-10 transceivers on a card instead of only 1-2. Several of the X2 members are also key contributors to the 300-pin transponder and XFP MSAs.
5. What are the specific benefits of X2?
- A. Versatility:
X2 has a versatile side rail system which allows X2 to support a variety of heat sink heights and mounting configurations with a single rail part number. Second generation 10 Gb/s applications show a lot more thermal variability than the more uniform "high end" switches that use XENPAK. So easily adapting the height (and direction) of the heat sink fins is a key benefit, With X2 this is possible with no changes or differences in the rail system, i.e., the same rail part number can be used for all heat sink heights and mounting configurations.
X2 allows uninterrupted airflow across the full area of the top of the module, for all height variants. X2 allows all the height in the vertical direction to be used for heat-sink fins, the side rail system does not consume any fin height. This is especially important for applications such as PCI which need every bit of cooling available in terms of fin height and need to achieve a clean entry and exit of airflow for the module in a side-to-side or front-to-back direction.
X2 to use the same "EMI solution as XENPAK, i.e., a flat "bezel" with a large overlap on the system front panel, and an EMI gasket in between the two. Our XENPAK experience confirms the obvious, that a large gasket overlap = a good EMI seal.
D. Port Density:
Physically X2 allows the key goal of "8 ports per line card" to be achieved. In reality port densities are likely to be dominated by thermal and EMI performance. X2's superior approach in these areas should allow real customer designs to support 8 or more ports per line card.
There is a strong support from key optical vendors and their respective customers for the X2 package. The X2 group contains many of today's top tier 10 Gigabit vendors. This means that working, standards compliant X2 parts are likely to be available, in volume, from multiple sources, in line with market needs (mid 03).
6. Will X2 and XPAK merge?
There has been discussion of an X2/XPAK merge. We support the development of a broad-based standard that would benefit our customers.
"We are very reluctant to enter into a point-to-point comparison article, which would continue to fan the flames of war," Gary Wiseman, director of marketing for Intel's optical platform group, told Light Reading today. "Our position is that the two should be merged," Wiseman says. "The technical differences are minor. The customer base has spoken -- it doesn't want two MSAs."
Of course, there are two ways of arriving at a single MSA -- merge them, or let one of them die. And the customer base might be hard pressed to decide which MSA should survive if Intel won't say why XPAK is better than X2, in as much detail as Agilent.
All the same, a merger sounds more likely right now. Intel has been working behind the scenes to try and reconcile the two groups from the beginning, Wiseman claims. In the early days of the MSAs, before they were made public, the two groups met to try and come to an agreement -- but failed. Now that both MSAs are more mature, Wiseman believes that they should try again to reach a compromise.
"We'll do whatever it takes," he proclaims. But clearly there must be some compromises that Intel isn't willing to make, otherwise the agreement would have been resolved first time 'round.
Agilent says merely "we support the development of a broad-based standard that would benefit our customers," adding that, "while discussions are ongoing, the two groups have yet to agree on specific technical areas such as versatility, thermal, EMI, and density."
Agilent, may have a vested interest in keeping the two MSAs separate, according to industry watchers. It is possible that it's got too much invested in the design to compromise even on small technical differences. For instance, it will want to keep the electrical connector the same, to allow customers of its Xenpak modules to upgrade to X2, which is essentially a smaller version.
Some have suggested that Agere's recent decision to exit the opto business may yet push Agilent into a compromise, however, notwithstanding the fact that the X2 MSA has considerable industry support from other module makers.
According to Intel's Wiseman, both groups will be meeting in the near future to tentatively discuss, again, the possibility of merging the standardization efforts.
Just so long as Agilent doesn't invade France Telecom...
— Pauline Rigby, Senior Editor, Light Reading