Routing switchers are basically an evolutionary line of products that can be traced back at least as far as patch panels. For about two decades, little changed other than rack density, signal quality specifications and warranty offerings. However, in the last 10 years, much has changed in keeping with the function that routing performs in the modern facility and the evolution of signal types.
The birth of digital routing
When digital routing was first introduced, it was parallel digital. If you never saw it, you might think I am out of my mind, but it consisted of 25-pin connectors with tightly controlled twisted pairs, all accurately matched in length. The distance the cables could run was limited. The Holy Grail was, of course, to simplify the cabling and deliver higher bandwidth and longer cable runs.
One engineer at a large European organization once told me that component serial digital routing was impractical, and the limits of physics would keep it from happening. That same year, a European manufacturer began delivering 270Mb/s serial routing that was pricey, but quite feasible for those with deep pockets, alongside its parallel digital products for 143Mb/s and 177Mb/s composite digital video. There were even production switchers with parallel digital inputs, which presented a real cabling challenge.
Video routing today
Today, we find a similar dynamic going on in the marketplace. High definition is barely out of its infancy as a consumer distribution medium. However, display manufacturers and professional equipment suppliers have cooked up 1080p60 hardware, despite the fact that for years, many experts have said that 1080p would never see a market in consumer space.
Well, those who said such (I could be rightfully accused) may find themselves quite wrong. Just as the 270Mb/s router was once considered barely an oddity, now we hear routing manufacturers touting the fact that they can handle bandwidths adequate to carry 1080p60 of 3Gb/s. 1080i60 requires the full capacity of SMPTE 292M (1.485Gb/s for 10 bits). Progressive scanning will require twice the pixels, or approximately 3Gb/s.
SMPTE has been working on standardizing a scaled-up version of SMPTE 292M specifically to accommodate 1080p60 and other future high-bandwidth connections. Twice the bandwidth on the same medium means that the distance will be more limited.
Some in the electronic cinema community feel that even this is not sufficient to handle more bit accuracy and 4:4:4 sampling, which they feel is a necessity in theatrical production and release. In rough calculation, if 1080p24 were sampled as 4:4:4 and 14 bits of depth, it would require 3.3Gb/s.
For electronic projection, the push is for 2k × 4k, which is four times as many pixels per second as 1920 × 1080. Thus, it is entirely possible the routing bandwidth envelope will be pushed a lot further in the future as applications demand infrastructure services that routing can deliver at even higher bandwidths.
The growth of bandwidth
The bandwidth is but one of several dimensions of the geometric space that routing occupies. As applications and facility sizes have evolved, routing has gotten even larger.
Last year, a single-level, embedded audio routing system was installed in an IPTV distribution facility, which was 2048 × 2048, resulting in approximately 4.2 million crosspoints in one level. Such massive needs are quite unusual, but routing systems of 512 × 512 and larger are commonplace today. With monitor walls having as many as 256 inputs, some facilities have moved monitoring paths to a separate level to control the creep of crosspoint and I/O count. The reason is quite simple: Doubling the I/O of a system moves the crosspoint count by a factor of four and the price by approximately a factor of three.
A second technique to control this growth is to use a distributed island of routing and pathfinding to allow signals to flow across a larger fabric. Just as a telecommunications facility is designed to deliver a practical pattern of use without allowing every crosspoint to be in use, a video facility assumes that statistically such a case is highly improbable and not supported.
By carefully analyzing the possibilities and organizing logical connections in smaller blocks, one can significantly reduce the size of a router without diminishing the usefulness of the system. For instance, ASI and SMPTE 259M both can pass through a monolithic router. Let's say you need 64 × 64 of each. You could put both in one monolithic 128 × 128 router or keep them separate because the signal types are incompatible anyway. A connection between an ASI input and an SMPTE 259M output is both illogical and improbable, and it is technically unlikely to be useful unless the receiving device automatically senses a choice of baseband and compressed video.
MTBF's effect on system reliability
There are other reasons to carefully consider how to implement large systems. One critical issue is MTBF and its effect on system reliability. Some manufacturers provide redundant crosspoints as a method of reducing the system MTBF. Others have provided methods of sensing the failure of an input signal and automatically replacing any instance of the failed source with a second copy on another, presumably operational, input. This can be attractive, but redundant crosspoints in a large router can be quite expensive. It is possible that a well-crafted distributed routing installation would be more reliable than a larger monolithic router with redundant crosspoints.
The future of video routing
Although this article deals with video routing, it is important that the idea of distributed routing has reached a logical extension in a product that was introduced and well-received at NAB2006. The product coupled TDM routing systems of modest size (128 I/Os) into a fabric router, which can be physically disbursed to different parts of a facility, putting the connections near the connected devices.
This could be powerful in a large facility, where long cable runs to a central plant router are often difficult. The product was intended for analog and digital audio, RS-422 and time code — not video — but if this concept extended into the video domain, it would have a wide market impact.
John Luff is the senior vice president of business development for AZCAR.