Stepping Up To 10G

Enterprise-class router unit shipments grew 14 percent in 2005, accounting for $3.3 billion in revenue, Infonetics Research reported. The market researcher forecast that annual unit shipments would grow another 41 percent by 2009 to $4.5 billion. Infonetics noted that while 10G is the fastest-growing segment, the bulk of the growth is still primarily in Gigabit Ethernet. Annual revenue for 10-Gigabit Ethernet switch port shipments topped $1 billion in 2006, IDC reported.

While 10-Gigabit Ethernet is still expensive compared to Gigabit Ethernet, prices are falling. The price of 10-Gigabit Ethernet fiber ports have dropped to about seven to eight times the price of Gigabit. For some vendors, it is trending even lower, approaching $1,000 per port. Contributing to the decline has been the development of a 10GbaseT standard for copper. According to SMC, if 10G Ethernet over fiber was $3,000 per port, then 10G Ethernet over copper cable would be about $1,000 per port.

According to Nelson Murga, product manager at networking vendor Ixia, copper is preferable to fiber because it takes advantage of existing infrastructure and is much easier to run. Fiber has more complexity, which may be a bit daunting. However, the physical distance limitations mean 10GbaseT over copper will more likely be used to connect data centers over short distances, rather than connecting large campuses.

A CAT6 cable can go up to 55 meters, while CAT6a and CAT7 can extend to 100 meters. A regular CAT5 cable is capable of only 30 meters. In contrast, Gigabit Ethernet over fiber extends 5 kilometers, and 10G Ethernet over fiber stretches to 40 kilometers. Vendors also are developing proprietary optics to increase the distance to 100 kilometers and further.

While copper is bringing the price down, there are other options to a 10G upgrade. If the customer does not need 10G, but Gigabit Ethernet is just not enough, one option is to bond several single Gigabit Ethernet networks together to create a larger pipe. This is a more flexible solution, especially if the customer will not be needing 10G for a long time. By the time the customer outgrows the aggregate trunk, the price for 10G would likely have fallen further and there may even be other options available.

Selling The Solution
When it comes to selling network upgrades, Labatt-Simon has an odd complaint. He said that the sales cycle is often "too short." Clients are reactive, not proactive, for they think about upgrades after they've already deployed an application on to the network, he said. At this point, the client is interested in a quick solution to get everything working, and the network assessment is rushed in order to get to the upgrade. There's no time to properly rearchitect the network.

Properly done, the assessment should last at least two to three weeks, and implementation should take anywhere from a little less than a month to a full year, depending on the network's complexity, he said. About 40 percent of the solution providers concurred in the recent CRN poll, saying the typical sales cycle is between two and four months.

Solution providers look at the current network utilization and determine the network's current usage loads and future needs in terms of applications and user size. NWN creates simulations based on the customer's network usage to discover patterns that can be used in capacity planning, Ludwig said.

The data center and backbone network utilization growth will provide a reasonable benchmark for when it will cross the 10G threshold. Administrators generally don't want to operate the network at more than 50 percent utilization because spikes and peak traffic can cause performance issues when the load creeps up.

Next: Room For New Vendors