Remember MRVC's Announcement regarding a Family of SFP Transceivers!!!!!
Small is Beautiful: The Inexorable Drive to Small, Hot-Pluggable, Uncooled Transceivers
Wary and cash-strapped equipment builders are not keen to jump on Small Form Factor Pluggable (SFP) transceivers until the price, technology and market are right. But someone has to jump first, and jump they will.
November 15, 2002
By John Devaney, CIR Managing Editor
That the demand for bandwidth has continued to grow, despite the economic slowdown, is undisputed. Meanwhile, "The one thing they ain't making any more of on this Earth is room," to paraphrase a recent episode of The Sopranos. Net result-port counts and line speeds can only keep on growing, while rack space isn't going anywhere fast.
Systems vendors, particularly in the Storage Area Networks (SAN) business, are fully aware of these pressures and are keen to keep up, by providing faster equipment with higher port densities. And next on the food chain, the makers of transceivers are eager to supply faster devices with a smaller footprint and lower power consumption.
Unfortunately, the road to the ideal transceiver is not a smooth one. Supplanting a well-entrenched technology like the Gigabit Interface Converter (GBIC) means competing on cost, as well specifications-not easy when GBIC manufacture is relatively mature and its market well-established and providing an economy of scale.
REACH FOR THE MARKET
Nonetheless, systems makers want smaller form factors and are phasing them in where there is no other way of achieving the port counts needed, and where the components meet specifications. In fact, one of the biggest complaints is that, whereas equipment makers have the luxury of choice between GBICs and "ordinary" Small Form Factor transceivers (SFFs) with a variety of transmission distance capabilities, Small Form Factor Pluggable transceivers (SFPs) are not available for transmission distances of 40 km and beyond.
One systems contact tells CIR: "We've been shipping 15, 40 and 80 km Gigabit Ethernet [GigE] for quite some time, so a new form factor that only serves one distance is really not a solution that's viable for us. [It means that] you need an Avalanche Photodiode [APD] for 40 km and beyond."
Of course, an APD adds to the total cost and takes up precious space on the line card, so the advantages of moving to SFPs might easily be outweighed. As a result, CIR has to conclude that GBICs will be around for some time to come and will even continue to dominate the GigE transceiver market beyond 2006.
The markets where longer reach has never been a significant selling point are Storage Area Networks (SANs) in general and Fibre Channel in particular. As a result, the recent shift from 1 Gbps to 2 Gbps as the dominant SAN communication speed was grabbed as an opportunity to move from GBIC to SFP technology. In fact, at a recent SAN conference, there was nary a GBIC to be found, although switch and server companies still support their 1 Gbps gear with GBICs.
The main consequence is that, although the GigE market for SFPs will build from just over $1 million in 2002, the Fibre Channel market for 2 Gbps SFPs is already worth more than $160 million. The experience gained in making and selling 2 Gbps SFPs for the SAN market can only help drive the price down and the transmission distances up for GigE SFPs.
There are other market opportunities that systems makers have identified for Gigabit SFPs, both in Ethernet and SONET. One contact tells CIR: "We use SFP on Coarse Wavelength Division Multiplexing [CWDM]-CWDM is fundamentally pluggable." While another states: "If you're doing metro core connectivity, you may need density and you may not need the distance, so it might make sense to go with the smaller form factor."
TAKING IT TO ANOTHER LEVEL
Evolution of 10 Gbps systems is some way behind lower data rates. In fact, it is reckoned that simply achieving industry consensus on a 10 Gbps small form factor Multi-Source Agreement (MSA) will take some time. At the moment, the main contender is the "does exactly what it says on the tin"-labeled 10-Gigabit Small Form Factor Pluggable (XFP) transceiver.
While some of CIR's contacts are attracted to, at least, the concept of the XFP because of its small size, its prospects are not so hot right now because of two major issues. First, it is difficult to design boards that carry 10 Gbps serial signals across them, along with problems of electromagnetic shielding, power supply and heat sinking. And second, the XFP MSA was only formally announced recently, at the Optical Fiber Communications Conference in March 2002, and only preliminary testable samples are available.
Again, the SAN industry is keeping a close eye on developments. A typical quote from CIR contacts in the field is: "Other things being equal, the XFP would be our form factor of choice." Also in its favor, the existing 10-Gigabit Pluggable MSA (XENPAK) is already considered to be too large and needs a special hole to be cut in the circuit board.
As a result of all this uncertainty, some systems builders are hedging their bets, sitting on all the MSA committees and waiting to see what plays out. Whatever it turns out to be, it will certainly be a faster, denser, passively-cooled future.
This article is based on CIR research within the Optical Components Advisory Program. For more information regarding this and other services, please contact John Jackson, Vice President of Sales, via email john.jackson@cir-inc.com |
