Showing posts with label Ethernet over copper. Show all posts
Showing posts with label Ethernet over copper. Show all posts

Tuesday, August 7, 2007

Private Line, Ethernet Might be Complementary

One of the confounding thing about "public network" services and platforms is that although many new services logically should displace older services, quite often they do so only in part, acting mostly as a brake on the growth of the legacy services, but not displacing them.

Ethernet and IP, for example, "logically" should replace older private line services based on SONET, SDH or optical carrier. Ethernet offers vastly lower price-per-bit performance and is transparent to the connectionless nature of IP. SONET, SDH and optical carrier can be made to encapsulate IP packets, but at the risk of additional overhead, cost and payload efficiency.

In fact, as IP-based broadband services proliferate over wired and wireless networks, one logically expects that older connection-oriented transport protocols will wither. But nothing in public networking ever seems to work so linearly. Oddly enough, as real time apps start to drive broadband services, connection-oriented transport has appeal, as that's what such protocols were created to do.

Over the past several years there also has been much emphasis on the role of wireless backhaul in driving new demand for private line capacity. Which might strike you as odd, given the relatively small percentage of total private line sales that particular application represents. Of course, there are other forces at work.

Though it clearly is broadband demand that is driving wireless backhaul demand, that demand is spread across traditional private line, Ethernet over copper and optical connections.

"Private line emulation" over optical or metallic media, for example, often makes sense. So does encapsulation of connection-oriented traffic inside a connectionless transport. Though "converged networks" are the future, today it often makes sense to add high capacity connectionless bandwidth for 3G and 4G services, but leave the connection-oriented voice on a separate logical network.

"Private line" sales can grow even as IP bandwidth grows in the backhaul application because a huge existing voice revenue stream has to be supported as incremental broadband apps using IP are layered on. Still, wireless backhaul is a fraction of total private line sales.

So why the buzz? Volume. A single sale to a wireless network provider can involve thousands of sites. A service provider obviously can make a lot more money selling one customer thousands of T1s or hundreds of optical carrier or Ethernet links, rather than thousands of customers single T1s.

Then there is the matter of urgency: wireless carriers have an immediate need that won't wait, and have to put up hundreds to thousands of links at a time. Wireless backhaul is really important to sellers because a handful of buyers represent such enormous volume.

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...