Optimizing Supply Chains for Online Video
A recent conversation with a white paper client took me back to a project from almost 20 years ago, when I was working side-by-side with supply chain experts at a well-known global consultancy.
“I have a background in agriculture,” says DDVTech founder and CEO Alfred Dijs, a bit modestly, as he made a career of working with several multinational companies in the Netherlands before setting his sites on the streaming industry.
Dijs was walking through the analogy of how a maturing streaming market could learn optimization lessons from the lower-tech transportation industry, an industry with technological advancements that have been leveraged to maintain a top spot in overall global exports.
“There are three game changers in the freight industry,” says Dijs when I ask about the mechanisms surrounding the Aalsmeer Flower Auction, which generates sales of 20 million flowers per week, to a tune of almost $2.5 billion per year.
“The first is standardization,” he says. “Streaming is moving towards standardization in the same way that Aalsmeer displays flowers in carts that can be easily loaded on to planes for delivery all over Europe.
“But standardization only works when the other two parts come together: independent networks working together and the availability of real-time stats,” he says. “The tracking and tracing of goods flow, directed by a set of uniform processes, is what allows cost efficiencies to be gained.”
Dijs uses the term “networks” in the traditional pre-internet sense: a group of people tasked with a particular job, be it brokers, longshoremen, or even customs officials. Each has a set of standards by which they operate, and each network is continually optimized to add incremental efficiencies on a monthly or yearly basis until some disruptive technology comes along to generate an orders-of-magnitude efficiency.
Streaming’s not yet through its standardization process, as we’re constantly packing and repacking different codecs and encoding approaches into the various container formats.
Those with a sharp intellect, however, will understand that data networks are built on the same premise as road systems, meaning that our packet collisions and timed exits or entries on to the Information Superhighway are key to success in our current, unicast-based environment.
In other words, we already have the networks designed for us, and we merely need to apply those networks to standardization, not the other way around. We see a practical example of this in the move to HTTP streaming: After years of trying to create new protocols to add to networks, we’re finally mature enough to think about how our shipping containers fit on the road with other traffic rather than trying to build a whole new highway system.
Model train buffs might think of it this way: You can’t put an N-scale model train on an HO-scale track, no matter how hard you try. For the rest of us, consider thinking of streaming less as broadcast and much more as transportation of goods.
Just because we are moving toward scale doesn’t necessarily mean that we’ll actually find the economies that come with scale. For certain, we do see some efficiencies already. For instance, the cost of bandwidth drops on a per-Mbps rate if one buys a larger data pipe, but only because of the efficiencies in the telecom and multisystem operator (MSO) networks.
On the CDN front, we assume that CDNs also generate real economies of scale when it comes to gigabytes delivered, as they offer corresponding price drops if larger storage and delivery contracts are signed. Yet the nature of streaming -- crossing network borders just like ships cross between continents -- means that the true efficiencies lie not in the ocean but in the communication needed to guarantee smooth and efficient operations when the ship reaches shore, whether that shore be the Land of Apple or the Island of Google or any other yahoo’s domain.
This article appears in the December 2013 issue of Streaming Media magazine as "Optimizing Streaming Supply Chains."
Companies and Suppliers Mentioned