Do copper's cost advantages at 1-Gbit/sec hold true at 10-Gbit speeds?
BY LISA HUFF, DISCERNING ANALYTICS
For networks running Gigabit Ethernet, it is a no-brainer to use Category 5e or Category 6 cabling with low-cost copper switches for connections 100 meters or shorter because they are reliable and cost about 40 percent less per port than short-wavelength optical switches. But for 10-Gbit Ethernet, users must consider other factors before choosing between copper and fiber.
While 10GBase-T ports are now available on switches from top vendors, is it really more cost-effective to use these instead of 10GBase-SR, 10GBase-CX4 or 10GBase-CR (SFP+ direct-attach copper)? It will depend on what your network looks like and how well your data center electrical-power and cooling structure are designed.
10GBase-CX4 is really a legacy product and is only available on a limited number of switches, so you may want to rule it out right away. If you are looking for higher density but you cannot support too many higher-power devices, I suggest you opt for 10GBase-SR because it has the lowest power consumption-usually less than 1 Watt/port. also, the useful life of laser-optimized multimode fiber is longer than that of the twinaxial cable used in CX4. Fiber is also smaller, so it will not take as much space, or block cooling airflow if installed under a raised floor.
If you do not have a power or density problem and can make do with just a few 10-Gbit ports for a short distance, you may choose to use 10GBase-CR, which will cost about $615/port. If you don't have a power or density issue and you need to go longer distances than about 10 meters, you will still need to use 10GBase-SR. and if you need a reach of more than 300 meters you will either need to: install OM4 multimode fiber-optic cable (which will get you up to 600 meters in some cases) to use with your SR devices; or look at 10GBase-LR modules at about $2,500/port, which will cost you nearly twice as much as the $1,300/port SR transceivers. If your reach is less than 100 meters and you can afford higher power, but you need the density, 10GBase-T at less than $500/port may be your solution.
If you have a mix of these requirements, you may want to make sure you opt for an SFP-based switch so you can mix long and short reaches, as well as mixing copper and fiber modules/cables, for maximum flexibility. (See sidebar.)
While the 10-Gbit Ethernet variants just described represent the greatest volume of data center applications, three other "flavors" of 10-Gbit Ethernet are worth discussing as well. They are ER, LX4 and LRM.
10GBase-ER runs over singlemode fiber for up to 40 kilometers. a few service providers may choose to use ER, but the vast majority of them opt for wavelength-division multiplexing (WDM) through the Optical Transport Network. Some private networks may have a need for this variant as well, which is what prompted the IEEE to include it in the standard. 10GBase-ER transceivers are priced out of typical budgets for the average enterprise, at more than $4,000.
10GBase-LX4 was originally introduced to address the installed base of FDDI-grade multimode fiber (MMF), but it can also be used with higher grades of laser-optimized multimode fiber (LOMMF) as well as singlemode fiber. It uses coarse WDM (CWDM) to send four wavelengths (thus the X4) between 1269 and 1355 nm running at data rates of 2.5 Gbits/sec each. LX4 modules are available in both X2 and XENPaK form factors, and from at least a couple of sources. Because these devices have four lasers and four detectors with their associated electronics, they cost appreciably more than either the SR or LR transceivers. The module alone retails for about $2,000, which means the per-port cost is probably close to $2,500. Users, however, most likely would not fill an entire switch with these modules but rather would employ them as needed in order to take advantage of existing installed fiber.
In response to the high-priced LX4, the 10GBase-LRM variant was developed. It was enabled by chip companies using electronic dispersion compensation (EDC), which could extend the reach of 10-Gbit multimode links. It took some time to develop the standard and as a consequence, LX4 took much of the market for which LRM was intended. However, once products were released and supported by top-tier transceiver manufacturers avago, Finisar and Opnext, it has taken much of the business away from LX4. LRM modules are now available in SFP+ packages as well, which indicates that the vendors believe there will be an ongoing market for them.
A note of caution: If you intend to use either LX4 or LRM modules, make sure the modules are in place at both ends of your channel. Otherwise, the channel will not work.
So what is the bottom line? assess the needs in your data center, and the rest of your network, and plan in advance to maximize the cost-effectiveness of your 10-Gbit networks.
LISA HUFF, CDCP is principal analyst with Discerning analytics LLC (www.discerninganalytics.com) and also chief technology analyst at datacenterstocks.com. This article is a compilation of several posts to her blog, opticalcomponents.blogspot.com. Lisa has spent more than 20 years in the communications business, in engineering and management roles. She has contributed to several of our Web seminars addressing technical and financial aspects of 10-Gbit/sec transmission.
The rise of SFP+
The rise of the SFP+ module is interesting. For those who have been in the industry for what might feel like 100 years, but is really just about 25 years, the one connector interface that has not changed much is the 8-position, 8-contact (8P8C) RJ-45. While improvements have been made to the connector by adding compensation for the 3-6 split pair, the connector itself has remained intact. Contrastingly, optical connectors for data communications applications have changed several times-ST to SC to MT-RJ to LC. They finally seem to have settled on the LC, and perhaps on a transceiver form factor: the SFP+.
The SFP was originally introduced at 1-Gbit/sec data rate, was used for 2- and 4-Gbit/sec and with slight improvements has become the SFP+ and the dominant form factor now used for 10 Gbits/sec. The SFP+ is in the process of getting some slight improvements again and promises to make it all the way to 32-Gbit/sec data transmission. That will be an impressive six generations of upgrades. How has the SFP+ done it? The InterNational Committee for Information Technology Standards (INCITS; www.incits.org) T11 Committee's Fibre Channel Physical Layer 5 (FC-PI-5) standard was ratified in September 2010. It specifies 16-Gbit/sec Fibre Channel. Meanwhile, the top transceiver manufacturers have been demonstrating prestandard 16-Gbit SFP+ devices.
Despite the common belief that short-wavelength vertical-cavity surface-emitting lasers (VCSELs) exhibited unstable modulation at data rates greater than 10 Gbits/sec, at least two vendors-avago and Finisar-have addressed the modulation-stability concern. Both suppliers say the engineering needed to assure the stability can be accomplished without adding significant cost to the modules. In fact, both also believe they can further push the capability of the SFP+ to 32-Gbits/sec through electronic dispersion compensation (EDC). This would allow its use in the next generation of networking data rates-32 Gbits/sec for Fibre Channel; 20 and 25 Gbits/sec for Ethernet and InfiniBand, respectively.
SFP+ is now also being employed for long-wavelength optical transmission. The small footprint of SFP+ has always made it challenging for the components needed to drive signals long distances, because the lasers within those devices require cooling. Fitting all that circuitry into an SFP+ interface went unachieved until Opnext showcased its 10-kilometer 16-Gbit Fibre Channel SFP+ devices in early 2010.
These developments involving SFP+ are important in the realm of optical data-communications networking for three reasons. 1) They show that Fibre Channel is not dead. 2) The optical connector and form-factor "wars" seem to have subsided to allow transceiver manufacturers and optical-component vendors to focus on cooperation instead of positioning. 3) These developments will impact the paths that other networking technologies are taking. Specifically, Ethernet and InfiniBand are using parallel optics for data rates above 10 Gbits/sec. Will they ever switch back to serial (single-lane) transmission?