Experts expect the 10 Gigabitt Ethernet (10Gig-E) standard to be completed by March 2002. One of the first questions to ask is, why should we need 10Gig-E when 1Gig-E was only standardized a couple of years ago? Because our desire to move data is almost insatiable—the more we move, the more we want to move.
(One)Gig-E sales reached $1.6 billion in 1999, according to International Data Corp. These sales are motivated by the low price of 10/100/1,000Base-T cards, which offer Ethernet, Fast Ethernet, and Gigabit Ethernet over Category 5 copper cable. Some cards cost under $200.
While a February 2001 survey by Network World concluded that only 3 percent of total LAN installations are running Gig-Es, it also showed that 44 percent of backbones run Gigabit.
Gigabit runs on twice as many backbones as either fiber distributed data interface (FDDI) or asynchronous transfer mode (ATM), the next-closest competitors. More crucially, the Ethernet market share is 96 percent, so 10Gig-E is the next-logical speed, especially for congested 1Gig-E networks! Some networks can use the 10Gig-E option as soon as it becomes available, and some vendors are already providing pre-standard products.
Beware: Although the draft standard discussed below appears close to final form, it still may change, so consumers should not rely upon it until it is finalized as is expected early next year.
10Gig-E has two important improvements to offer compared to its predecessor, Gigabit Ethernet. First, and most obvious, is its higher speed. Secondly, longer distance appropriate for wide area networks (WANs), not just for local area networks (LANs) and metropolitan area networks (MANs).
In terms of telecommunication physics, this is counterintuitive, because normally speed and distance are inversely related, so higher speed would normally mean a shorter distance.
The capability to span greater distances results from industry experience with the Gigabit Ethernet, which showed the efficiency of the Ethernet frame in carrying data in MANs and WANs. Hence, it has received much attention in the 10-Gigabit standard creation.
Two crucial adjustments were needed to make these enhanced capabilities possible. First, it was decided from the outset that it will use all optical wiring with no copper wiring at all. Second, the standard will support full duplex only, so no collisions will occur and no Carrier Sensing Multiple Access with Collision Detection (CSMA/CD) scheme will be used as it was previously. Both of these are tradeoffs.
Installed optical fiber costs more than installed unshielded twisted pair (UTP) cable, and full duplex limits the number of communicating devices to only two. Nevertheless, these tradeoffs are accepted to achieve the desired mix of speed and distance.
WAN first, then LAN
Before a technology reaches its ultimate popularity, it must be standardized. Even as the Gig-E standard was being finalized in 1999, a study group was formed for 10Gig-E. This was formalized as the Institute of Electrical and Electronics Engineers (IEEE) 802.3ae group in the spring of 2000, and the final standard is expected in March of 2002.
Another heavily involved group is the 10 Gigabit Ethernet Alliance. This group, nicknamed 10GEA, was established to promote standards-based 10Gig-E technology and to encourage its use. More in-depth information is on the Web site at www.10gea.org/.
Since the major motivation for the standard came from vendors and carriers interested in long-distance applications, the first standards efforts concentrated on single-mode fiber interfaces and almost ignored (or proposed to postpone) the 850nm serial multimode fiber interfaces most common in LANs.
This seemed upside down because Ethernet itself originated in LANs. This less-expensive option was ultimately added to the set.
Gigabit standard distances
The table above of devices, media, and distances has been published on the 10GEA Web site. It shows the all-optical nature and the extended distances possible with the upcoming standard.
Concerning the table: Wide Wave Division Multiplexing (WWDM), a newer technology that allows light from several different laser sources to be sent through the same fiber at several different wavelengths, it differs from Dense Wave Division Multiplexing (DWDM), a more familiar technology.
100Gig-E is next
100Gig-E isn’t that far off, either! We can anticipate that whenever the 10Gig-E task force finishes its standard (or maybe before), work will immediately begin on the next standard. Not surprisingly, this will occur as we use our networks more often.
SHINN is principal of Compututor. He can be reached via e-mail at firstname.lastname@example.org.