Paving Paradise
by Lee H Goldberg

Don't it always seem to go
That you don't know what you've got 'till it's gone
They paved paradise and put up a parking lot
- Joni Mitchell, Big Yellow Taxi

The networking industry may be about to hit a hidden speed bump as the number of semiconductor companies actively involved with developing products to support SONET/SDH, PDH, and other TDM-based technologies can now be counted on one hand. You might wonder why this is a problem a bit more than a decade since the imminent death of TDM-based transport technologies was predicted by the net-heads, until you realize that they still comprise a significant portion of the Big Tubes and access infrastructure in use throughout the world. With the IP revolution having morphed into an evolution, the dominant life form in the datacom cosmos is the converged network where TDM and IP are finally interworking with a minimum of fuss. But without a critical mass of companies to support the TDM side of the equation, that evolution could slow to a crawl.

The last few years have seen most of the original players in SONET/SDH semiconductors either cutting back, or bowing out completely, from the market. I could talk about how Vitesse and Infineon seem to be simply flogging their existing parts or about Cypress' complete exit from the market, but the recent acquisition of Agere by LSI Logic stands out as a key event in this saga. I cannot read the minds of the LSI executives, but it's almost certain that they bought the remains of the once-mighty company that helped write the SONET specs for their storage technology assets and have little or no interest in the framers, mappers and network processors that represent the pinnacle of a 40+ year learning curve.

It's debatable as to whether LSI will allow the remains of the telecom division to limp along, try to sell off the assets, or simply plow it under, but it's almost certain that the past few years of buffeting, abuse and neglect have already destroyed the ranks of a once-powerful brain trust for a critical technology...to the point where it's uncertain whether a critical mass of expertise can ever be re-assembled. I suspect similar scenarios are playing themselves out at several other companies as the tech industry leadership's notoriously short attention span drives product development away from mature core businesses to pursue sexy new opportunities in wireless multimedia, EPON gateways, or some other Next Big Thing.

In a recent conversation, the CEO of a small semiconductor company said he thinks that this thinning of the herd we're seeing is because today's market is still being shaped by the 2001 telecom implosion. He feels that while the overall telecom industry is healthy again, with lots of new deployments and equipment sales, it's being supported by lots of chip makers and even some equipment makers who are still losing money for a variety of reasons we won't go into here. Between this and the recent flurry of stock scandals, corporate investors have begun to question companies' abilities to produce a return on their capital. This has usually led them to impose severe constraints on future R&D efforts, and much firmer requirements for short-term returns. In other words, the 5-7 year business cycle of telecom products is much less attractive to folks looking for the quick buck, causing most companies to shy away from the next generation of framers, mappers, ADMs, and other devices that go into infrastructure equipment.

If that CEO's analysis is anywhere near correct, it would explain why I've seen so little action in the TDM semiconductor sector these days, despite the fact that SONET/SDH and PDH are critical elements of the converged networking products that are so hot right now. It also would explain why only a few companies like Transwitch, Galazar and, to a limited extent, AMCC are the only folks pushing ahead with the development of new SONET/SDH or EoS/PDH silicon after a 7-year nuclear winter. Most likely the gradual thinning of the ranks we've seen should slow while there are still a few companies capable of developing the next generation of carrier-class SONET/SDH and EoS devices we'll need to continue our migration towards an all-IP future. Nevertheless, I still worry about whether financial pressures will induce a violent shakeout that takes us below a critical mass of TDM know-how before we're ready to let it go.

Given the short-sighted perspective of our industry, we may not know what we had 'till it's gone.

Comments? Questions? Thoughts on the future of SONET/SDH you want to share? Write me at lhg at en-genius.net.