Electronic Design

Communications: Optical

Staggering Optics Market Reshaping For 2005

It looks like it's going to be another down year for optical, so the angst continues. In fact, some say that the decline will continue through 2004, with a return to growth in the following years. The field of fiber optics grew to unexpected proportions in 1999 and 2000, but it began to crash with the telecommunications meltdown in 2001. Today it is just a shadow of its former self, but it's still around and waiting for the rebound that must ultimately occur.

In 2003, the optical industry will continue the consolidation that began last year. Both large and small companies are going out of business, selling off or dropping their optical product lines, and implementing all manner of mergers and acquisitions. Some big companies with cash hoards are taking advantage of the downturn to build an optical business at bargain prices. What will result is a smaller but healthier optical industry. The good news is that R&D has continued in many companies, although on a smaller and slower scale. These companies will be ready with hot new products when the upswing begins. The bad news is that the fiber-optic business may not have hit bottom yet. Given that cheery news, look for 2004 and beyond for recovery.

The optical industry almost totally depends on the telecommunications industry, which continues in a hunkered-down mode. With all of its financial problems, overcapacity in its long-haul networks, and lagging sales of services, the industry is simply in no position to buy new optical networks or equipment. Instead, the industry is in a "make do" mood. Companies are fine-tuning their current networks with minor equipment upgrades and software "tweaks." The result has been improved utilization efficiency from normal levels of less than 50% to as high as 70% in some cases. This may hamper peak load demand, but right now it's manageable in light of the economic conditions.

There has been some growth in other sectors using fiber optics, including metro networks, storage-area networks (SANs), and 1-Gigabit Ethernet (1GE) and 10-Gigabit Ethernet (10GE). In fact, 1GE and 10GE are being implemented in local-area networks (LANs) and metropolitan-area networks (MANs). New technologies such as Internet Small Computer Systems Interface (iSCSI or I-scuzzy), Resilient Packet Ring (RPR), 10-Gbit/s line speed network processing units (NPUs), microelectromechanical systems (MEMS), and tunable lasers are becoming available. This has spurred on some growth given the present downturn.

Gloomy reports aside, the optical networking space is not dead. It will return just about the time the industry is ready for the heady products now being developed.

See associated figure.

>SLOW BUT SURE ADOPTION OF 10-GBIT/S SYSTEMS. There has been lots of talk about 10-Gbit/s systems for the past several years, as component manufacturers have announced and even delivered their 10-Gbit/s products. But few real OC-192 Sonet networks have yet to be installed. Most of the traffic for telecom and the Internet is still carried by OC-48 2.5-Gbit/s networks. But now 10-Gbit/s equipment is available, competitively priced and ready to go.

>NO 40-GBIT/S SYSTEMS YET. Work continues as the component manufacturers learn how to make OC-768 parts and systems. No complete system has been installed yet, but look for the first one in late 2003 or early 2004. Achieving 40 Gbits/s is not easy. CMOS doesn't quite cut it... yet. For now, indium-phosphide (InP) and silicon-germanium (SiGe) ICs and discretes are getting the job done. Fiber manufacturers are still dealing with the many types of dispersion that thwart long-haul 40-Gbit/s systems. Other companies are developing compensators, so it looks like we will have real live OC-768 and maybe even a 40-Gbit/s Ethernet eventually.

>MORE AND BETTER DWDM. Development continues on DWDM components and systems. Manufacturers now have splitters that can separate the channels with 12.5-GHz spacing. Systems with over 200 channels are now easily implemented. With data rates of 10 Gbits/s per channel, the aggregated data rate is now in the petabit range. While that is a useful development, many vendors are using DWDM to achieve high aggregate rates with slower multichannel delivery. This is expensive right now, but the new tunable lasers will make it de facto in the future for some systems.

>TUNABLE LASERS HIT THE MARKET AT LAST. It costs a fortune to implement and maintain a big DWDM system. With one laser per * and each laser selling for $4000 to $10,000, the cost of a 100-channel system is over the top, even without the cost of keeping spares on hand for repairs. Big bucks are involved. Tunable lasers have been in development for years, but we are now beginning to see some real practical and affordable products that can solve this problem. While we're not at the point where one laser can do the whole job, we can look for new systems with only three lasers.

>THE ONE WORD IS NOT "PLASTICS," IT'S MEMS. If you've never Dustin Hoffman's movie The Graduate, then you probably don't have a clue what I'm talking about. In any case, microelectromechanical systems (MEMS) are emerging from the R&D labs of the past few years to produce real products. These unbelievably small mechanical semiconductors are ideal for some optical applications. Specifically, MEMS mirrors are showing up in some of the newer optical switches and attenuators.

>NO AON OR PON ROLLOUT YET. Work continues on the all-optical network (AON) and the passive optical network (PON). While a few PONs have been installed in fiber-to-the-home systems, it's not widespread. Some regional Bell operating companies (RBOCs) and cable TV companies have extended their existing networks with PON fiber to the home to test its viability and performance. Success could lead to PON becoming the replacement of the current long-lived local loop.

>METRO NETWORKS WILL KEEP THE OPTICAL-FIBER field healthy until recovery. With long-haul fiber capacity at a maximum and continuing growth in Internet access subscribers, it has become evident that the bottleneck is clearly in the metropolitan access part of the system. New metropolitan-area networks (MANs) are being installed, many of which have the advantages of newer technologies.

>ETHERNET INCREASES ITS NETWORKING MONOPOLY. The new optical versions of Ethernet can produce data rates of 1 Gbit/s and 10 Gbits/s over significant distances. While these versions were designed to extend the LAN, many have found that they also work well in MANs. They cost less and can interface directly to LANs without the usual conversion from Sonet rings. Some are even advocating the use of 10GE in WANs, but it remains to be seen if it will provide the level of service most carriers demand.

>THE RESILIENT PACKET RING (RPR) STANDARD will be ratified, and systems will begin to emerge. The IEEE RPR working group is expected to ratify this standard, designated 802.17, this year. Some vendors are already offering systems. RPR is a media-access-control (MAC) protocol that's expected to be used with Sonet rings to provide more efficient access to shared media, especially in metro networks. It permits Sonet to aggregate bursty data, such as transmission-control protocol/Internet protocol (TCP/IP) and Ethernet, while providing traffic management, multiprotocol label switching (MPLS), and carrier-class performance.

>SIGNIFICANT GROWTH IS expected in SANs. Most SANs are already well entrenched in the larger organizations and use Fibre Channel (FC) networking equipment. A newer alternative is the Internet Small Computer Systems Interface (iSCSI). This block-oriented Internet Engineering Task Force (IETF) protocol uses Ethernet to implement SANs much less expensively over nearly unlimited distances and with IP security (IPsec) if needed. Look for Fibre Channel vendors to address current interoperability problems, implement a 10-Gbit/s version, and lower costs to remain competitive.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.