SPIE Membership Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
SPIE Photonics West 2019 | Call for Papers

2018 SPIE Optics + Photonics | Register Today



Print PageEmail PageView PDF

Optoelectronics & Communications

Will optical replace electronic packet switching?

Despite advances in the technology, all-optical packet switching will not be competitive with its conventional electronic counterpart.
20 May 2007, SPIE Newsroom. DOI: 10.1117/2.1200704.0687

Packet switches (or routers) are key hardware components in the Internet, directing information through the network from its source to its destination. Incoming data packets arrive at routers on optical-fiber transmission links and are converted to electrical form by optical-to-electronic (O/E) converters for processing. Similarly, outgoing data passes through E/O converters before traversing the next optical link in the network. An optical packet switch is similar to an electronic router but does not require O/E and E/O converters.1–3 Instead, the data remains in optical form throughout the network, including the switches. Figure 1 shows schematic diagrams of electrical and optical switches.

Figure 1. Schematic of electrical (upper diagram) and optical (lower diagram) packet switches. In both cases, the input/output (I/O) ports connect to line cards that contain a CPU, an interface, and a buffer. The line cards connect to the cross-connect switch fabric. An IP lookup table and a CPU control packet forwarding. In the electrical switch, all components are electronic except for the optical-to-electronic and electronic-to-optical converters at the I/O ports. In the optical switch, the input interfaces, buffers, and cross-connects are optical.

For a number of years, researchers (including the author) have argued that optical packet switching will become an essential part of the Internet as the size and capacity of the network expand. The approach has several key advantages, including higher bandwidth and data format independence. However, a major obstacle to development of practical hardware has been the lack of a suitable memory technology that can store high-bit-rate optical packets. In an electronic switch, electronic buffer memory provides storage when needed to avoid collisions between outgoing packets on the same output port.

In a recent paper,4 we reported a detailed study that compared the advantages of optical and electronic switching, with a focus on power dissipation. This key consideration has largely been ignored by the optical packet switching research community.1–3 However, the power consumption and heat dissipation associated with big routers are already major issues in the telecommunications industry5 and are likely to become even more problematic as routers become larger. Optical switching will never play a significant role in the Internet if it consumes more energy than electronic switching.

Our study included the two main physical-layer parts of a router: the buffer and the cross-connect switch. We used aggressive but plausible estimates of buffer and switch performance projected out to around 2020. Our analysis of buffer technologies was based on models of power dissipation in optical delay-line buffers and in high-capacity, rapidly reconfigurable optical cross-connects. We also looked at storage capacity in slow-light delay lines. In addition, we compared conventional fiber delay-line buffers with slow-light delay-line buffers using photonic crystals, resonator structures, and electromagnetically induced transparency.6–9 Finally, we examined the physical size of the hardware required. To give concrete examples, we considered packet switches with throughputs of 100Tb/s (1014 b/s) and 1Pb/s (1015 b/s).

We concluded that optical packet switching does not provide a magic solution to the problem of providing an ever-expanding switching capability in the Internet.4 Optical routing does not solve the energy problem, and neither conventional nor slow-light delay lines do away with the storage problem. In particular, planar integrated optical buffers (including slow-light buffers) occupy a larger chip area than their electronic equivalents, dissipate more power, and are limited in capacity to, at most, a few IP packets. Optical-fiber-based buffers have low power dissipation but are bulky. Optical components will play a role in large-scale cross-connects, but O/E/O converters will not be eliminated. Optics is well suited to transmission of data, and electronics is well suited to buffering and manipulation of data.

Note that unforseen technology breakthroughs could well change some of our conclusions. However, we hope that our work stimulates discussion and debate on the relative merits of electronic and optical packet switching. If this debate eventuates, it needs to consider the centrally important issue of energy consumption.

Rod Tucker
ARC Special Research Centre for Ultra-Broadband Information Networks,
University of Melbourne
Parkville, Australia

Rod Tucker is a Federation Fellow and Laureate Professor at the University of Melbourne. He is research director of the ARC Special Research Centre for Ultra-Broadband Information Networks.