* Vativ spokesman criticizes IEEE’s work
Last time, we saw the beginning of a debate over Vativ Technologies’ approach to 10 Gigabit Ethernet, which differs from the copper-based 10 Gigabit Ethernet standard under development by the IEEE.
Vativ Director of Marketing Steve Kotowski criticized the IEEE and wrote that a non-standard approach could be better:
“Vativ has taken a different approach than the standards bodies, since the standards bodies have failed the marketplace in their attempt to deliver a simple, low-cost 10GbE solution. Likewise, their efforts on 10GBase-T are likely to have the same result. Vativ’s solution is low-cost, easy to manage and most importantly, available now.”
IEEE 802.3an Task Force Chair Brad Booth, who is also Ethernet architect at Intel, countered:
“If Vativ felt so strongly about their solution, then why didn’t they approach 802.3 with a project to standardize this?”
Kotowski further attacked the IEEE’s 10GBase-T effort:
“We do not believe that 10GBase-T is a viable solution for data center applications. 10GBase-T is targeted for 100m reach with Cat 7 cables… It is being touted as a desktop solution. We do not believe that there is a viable market for 10GBase-T. First and foremost, since it requires new cabling installations, it cannot be used with existing wiring in the building.”
Booth shot back:
“10GBase-T has never been touted as a desktop solution. It was stated that because 10GBase-T uses structured cabling standards that end users could take it to the desktop if they wanted, but its target market is the data center (where 100m reaches are possible). As for the market viability, many system vendors support 10GBase-T, and they’re the ones that eventually buy the PHYs.”
There was one area where the two agreed – but even there they had differences. Kotowski wrote:
“We expect that 10GBase-T will not be practical or compatible with existing system designs due to the high power consumption of the devices (estimated to be [about] 15W per port in 65 nm CMOS technology). It is also not a scalable solution because it will never fit into an MSA module due to high power consumption. An equally important consideration is that due to the complexity of the architecture and the high power, silicon solutions will be large and require expensive packaging coupled with advanced thermal management. As a result, the overall packaged die cost will be prohibitively high for the foreseeable future.”
Booth said:
“I do agree that power will be a big issue, but 15W in 65nm exceeds most of the estimates. 1000Base-T was very high in power when it first came to the market, but in time that power number dropped substantially.”
