10GBase-T Musings

I’m thinking about doing some digging on 10GBase-T offerings. My initial thoughts are very scattered. The topic is intriguing because one hand 10GBase-T is a “no-brainer” ( it will be largely deployed some day) and on the other hand 10xBase-T has always been very difficult to execute on. It can become a “development money pit”. 10xBase-T is a very expensive development project which often requires the Silicon technology of tomorrow. Thus the first chip out is not necessarily an advantange. Standing up to cross the finish line may/will be difficult. 💡

Here is what I’m starting with:

  • Too much power dissipation (Pd) – Andrew@Nyquist estimates that offerings in 2011 will have manageable power. Linley wire concurs ( see below)

    • Here is a clip from Linley Wire May 2007

      By using 65nm process technology, Teranetics should be able to reduce power dissipation – it has promised to sample a 5W device by 2008. Although the first wave of 10GBase-T PHYs should ship in significant volumes, it will be the second generation of devices that ramps up to millions of 10GBase-T ports.

    • What does this mean?

      • translates to — the IC technology requires 2 more process/line_width reductions. ( or 2 more full tape-outs )
      • This comment is most likely for quad-port devices for switches. A single port device suitable for servers may be ok now or in next process point.
  • The development cost is already stratospheric and they’ve got to do another version ❗
    • What is the business case for CommSemi developments of over $100M? ( eg. SolarFlare ~$100M)
  • 10xBaseT mkt has historically started with NIC cards for Servers — but how many servers can really pump out more than a quad-GbE? This is a specialized market for now.
    • I’m sensitive to this ’cause I was once a product manager for GigE NICs 😦
    • The early 10GbE market is switch aggregation and Fibre is ok for now.
  • The developers are Solarflare, Teranetics, Aquantia, KeyEye, Plato Networks as well ast BRCM, MRVL, and VTSS (according to Linley Group report .. this is exec summary.)

One thought on “10GBase-T Musings”

Comments are closed.