How to Control Latency with Optical Systems
By Time is of the essence. That is especially true for applications like high frequency trading. Remember Wall Street’s May 2010 flash crash that sent the Dow down nearly 1,000 points within seconds?
As I wrote following that event, this incident illustrated that every second – or even millisecond – can make a significant difference, whether you’re talking about the quality of delay-sensitive traffic, the end user experience with cloud-based services, or the ability to trade fairly (or quickly halt trade, when needed).
Latency is also the theme of a recent article by Mannix O’Connor, technical marketing director of MRV Communications. In it, he mentioned that in the past few years, latency has been reduced by enhancing CPU cycles and speeds, the input and output speed of devices, network bandwidth rates, OS speeds, and software computation speeds. Latency introduced by the fiber optic networks over which the data travels is also an important consideration, he adds, noting that the maximum speed of light (300,000 km per second or 5 microseconds per kilometer) is a key constraint here.
Shortening the distance of the connection is one way to cope with that constraint, he reminds us, but where applications, fiber, and real estate exist is a gating factor. Using DWDM technology to create virtual pipes within the fiber can also help organizations get better network performance, he says. But while DWDM is expanding from 10gbps, to 100gbps starting early this year, and even up to 200gbps by the end of the year, not all 100G components have the same latency, he says.
“When moving to 100G it is important to understand the latency from the transmit side of the link to the receive side of the link,” he adds. “This measurement will show which 100G dispersion compensation components have the lowest latency.”
To contend with all that, O’Connor says, organizations can use latency-optimized optical systems. These systems should include dispersion compensation, which broadens the input signal as it travels through the fiber, he says, suggesting that the best solutions for this add latency of just 0.15 nanoseconds per kilometer. Selecting passive optical muxes with the lowest latency can also help, he says. And he adds that 100G transponders are available that have less than 5 nanoseconds of latency.
Edited by Maurice Nagle
