Monthly Archives: August 2008
SOA, Rhymes with Noah
by Leslie Ellis // August 25 2008
And now for a quick dip into the bubbling soup of acronyms within the software world, starting at the top. “SOA.” People say it as a word — rhymes with Noah.
“SOA” stands for “Service Oriented Architecture.” It’s the Big Picture for the software efforts of big companies. It’s especially enticing to companies wishing to untangle themselves from heavy, monolithic, single-vendor software systems.
Like the billing system, for instance. The historic grumble about cable billing systems goes like this: Ask for a change. Wait 18 months. Find a million dollars to pay for it.
That’s why you tend to hear of SOA when you’re with IT people. Here’s a usage example from a recent batch of notes: “We took a look at it and said, we need a SOA architecture, to let us to get time to market and productivity enhancements.”
Try this for fun: With a straight, calm face, suggest to anyone who works in cable IT that they’ll need to change out the billing system. Then try to find a way to share in the utter hilarity of the idea.
Here’s what SOA is: It’s tight, efficient little blocks of code, theoretically re-usable, with consistent passageways between them.
In practice, SOA is seeing that 60% of your care calls about digital video result in sending a refresh command to the box. Wouldn’t it be great if there was a way to let customers initiate the refresh themselves, by pressing seven on their phone, or asking online?
Pre-SOA, pinging a set-top required a care agent to initiate that activity, by accessing the headend components of what were then General Instrument and Scientific-Atlanta systems.
With SOA, pinging a set-top means abstracting that function into a chunk of code, then embedding that chunk of code into the other chunks of code that might need it — the IVR system for the phone; the self-care portal for the online query.
The catch: Those “theoretically re-usable” chunks of code. Say a “service,” as the chunks of code are called, moves into the domain of another “service” needing it. To the web care portal, in this example.
Oops. It only covers 80% of what the web portal needs. The other 20% either comes from an add-on, or, just as often, a total re-write.
Most of the larger cable MSOs are at least waist-deep in SOA, so it’s where your IT friends are headed. May they prosper.
This column originally appeared in the Platforms section of Multichannel News.
Advanced Advertising and SCTE 130
by Leslie Ellis // August 18 2008
Earlier this year, I tagged an advanced advertising standard, then known as SCTE DVS 629, as one of the “big things in cable tech” in 2008.
Since then, and to keep us all on our toes, the SCTE Digital Video Subcommittee changed the number of the standard, from DVS 629 to DVS 130. Not sure why. It reminds me of a funny friend’s funny child, who counts in this order: “One, two, skip a few, 99, 100.”
So: DVS 629 is now DVS 130. And DVS 130 is the sockets and sinew of addressable, targeted, interactive advertising, on cable.
On Aug. 4, the first four parts of the eight-part DVS 130 standard were approved by the SCTE’s Engineering Committee. That greenlights the vendor community to begin interpreting and building. (Many already have. Doing business in a standards-populated economy often means sprinting ahead, hoping your work becomes “the standard.”)
This week’s column attempts to untangle the context of SCTE 130. The actual nuts and bolts are a different translation. Consider: The four finished parts of the standard run across 429 pages.
If you’d rather fast-forward to the summary translation, it goes like this: DVS 130 creates the framework to pick, on the fly, which ad, of which length, to splice into a TV show — whether that show is linear, stored, or switched. It’s how to advertise the stackable washer/dryer to Condo Connie, and the lawnmower to Harry Homeowner.
For VOD, DVS 130 leapfrogs the “bookend” ads currently glued to the front and back of a video title. It adds things like replacement ads, pause ads , and telescoping capabilities.
Replacement ads do what the name implies: Splice a newer ad over the existing one. (It’s harder than it sounds, because it crosses industries. Not all program networks have the gear to put the necessary flag into the ad breaks of a TV title, to indicate when a cable ad substitution can happen.)
Pause ads pop something up when you decide it’s time to break for a ham sandwich.
Telescoping is the clickable thing that invites the interested viewer to see a longer, stored video.
As a framework, DVS 130 defines the language to be spoken between participating machinery, and what messages they’ll exchange. Likewise, it defines how to connect the machines doing the work of addressable and interactive advertising.
SCTE 130 doesn’t define how that targeting and campaign work should be done — that’s the job of innovation.
The Deeper Dive
If you haven’t observed a technical presentation on DVS 130, know going in that it’s pretty architectural. That means diagrams best absorbed by printing them out and staring at them. For a long time. With a clear head.
DVS 130 registers heavy on the jargon meter, too. Its official title: Digital Program Insertion — Advanced Advertising Interfaces. Its remaining four parts are expected to be approved this year. They’re juicier. They get at the raw materials of how to address ads to Condo Connie, based on what she approves you to know about her.
And yes, this whole thing will work best when all eight parts are done and in motion. In reality, that means this is a 2009-2010 thing, if everything goes well.
But until then, here’s the short version of the four approved parts of SCTE 130:
Part One, Advanced System Overview (16 pages) summarizes parts two through eight.
Part Two, Core Data Elements (77 pages), defines how to phrase XML (Extensible Markup Language) messages for addressability and interactivity.
Part Three, Ad Management Service (ADM) / Ad Decision Services (ADS, 246 pages), gets pretty dense. In essence, an ADM issues messages about what ads to place, An ADS figures out how to place them.
Part Three puts “real time” into the equation. It’s the “advanced” of advanced addressability. Today, it works like this: Ads are sold. Traffic schedules are built. At 4:00 p.m., those schedules get loaded into the ad insertion machines. If something needs to change after that, it better be important.
Part Four, Content Information Service (90 pages) is the keeper of the metadata about the ads and the video content they’ll run within.
Modularity and Scale
The fact that DVS 130 is chunked into eight parts illustrates one of its intents: To be modular in design. In premise, modularity attracts a wider supplier community. Plus, it lessens the risk of ganging stuff together that grows at different rates — scale matters.
Remember the first days of VOD gear, when storage and streaming worked in the same box? Storage grew faster. Decoupling happened.
Another handy consequence of SCTE 130 is the data it gathers — house by house, system by system, region by region, operator by operator — all the way upriver to the Canoe Ventures LLC, if so desired. (Yes, it’s an actual company now, with headquarters in the Chrysler Building, in Manhattan.)
When it all comes together, advanced advertising will send you the ads that are best for you, assuming you’re ok with it. Your viewing becomes collectable data, which gets ganged together with everybody else’s collectable data. Once the data is sufficiently smooshed and “anonymized,” cable advertisements become targetable and measurable.
That’s huge, for the people who live and work in cable advertising. It’s several giant steps towards being “more like Internet advertising.”
This platform originally ran in the Platforms section of Multichannel News.
The Secret Bandwidth of Addressable Advertising
by Leslie Ellis // August 11 2008
Last summer, while talking with a small gathering of muckety-mucks from a cable program network about how cable operators use bandwidth, I was asked what I knew about the “secret bandwidth.”
Secret bandwidth. News to me.
I asked for particulars. Turns out that this executive had learned from a cable operator contact that in addition to the digital shelf-space dedicated to standard and high definition linear TV, a special reserve also existed for advanced advertising.
I resisted the urge to tell the guy that he’d been tricked into some kind of digital snipe-hunt.
A year passed. Then, last week, a muckety-muck of the advanced advertising persuasion took a pause from his heuvos rancheros to ruminate over the bandwidth implications of addressable advertising, in high definition.
In order to send a 30-second spot that is, for whatever reason, more targeted than what’s already embedded in the linear video stream for that show, he said, operators will probably need to reserve some portion of their existing bandwidth.
Tactically, it goes like this: A typical, 6 MHz digital cable channel carries 10 to 12 linear video streams, in standard definition. To do addressable advertising, the substitute ads need some carriage room, too. Like three or four of those 10 to 12 streams.
Aha! The secret bandwidth.
Predictably, different addressable advertising vendors do this differently. Some borrow streams from within a mux, as described. Others ask for dedicated capacity — one to two 6 MHz channels, to carry the addressable ads.
(Proponents of the first method say that proponents of the second method introduce latency issues, because the set-top box has to physically re-tune to another multiplex, in order to grab and display the ads from the secondary or tertiary channel. Proponents of the second method say their way is a better use of bandwidth, because of statistical multiplexing gains.)
In HD, though, the matter gets more pronounced, because only two to three HD streams fit into that same 6 MHz digital cable channel — so where do the addressable ads go?
Three options, each with an increasing level of complexity. One: Dedicate additional digital channels to the needs of addressable ads. Two: Go faster on the move toward a unicast architecture — where each household gets its own video stream, and its own ads. Three: Find a way to use advanced compression — to mix MPEG-2 and MPEG-4 traffic in the same channel.
To those of you who, like me, were stumped chumps about the secret bandwidth — there it is. To those who already knew — my apologies for assuming a digital snipe.
POST-SCRIPT: Since this column ran, several readers, all of whom work for cable operators, wrote to let me know that switched digital video implementations can tuck in addressable ads without using tangible 6 MHz channels. If an advertisement is addressed in a forest, and no tangible bandwidth is used to carry it, is it still a secret? 😉
This column originally ran in the Platforms section of Multichannel News.
Upstream Bandwidth & Symmetry
by Leslie Ellis // August 04 2008
Recently, a childhood friend wrote to express his outrage at the U.S. cable industry, for “deliberately precluding people from getting symmetrical broadband speeds.” Several angry blogs bludgeon this topic, too. Salty opinions abound.
This week’s translation will attempt to explain (hopefully without damaging the friendship) why this particular accusation is incorrect — by reasons of regulation, physics, and present reality.
Let’s take it from the top. “Symmetric,” in a bandwidth sense, means the same amount of speed goes toward the computer as away from it. Most of today’s operators provide asymmetrical packages — so many Megabits per second toward the computer; fewer Megabits per second away from it.
This fits the pattern of most early and ensuing Web traffic. Your request for a web page, video stream, or a voice call is substantially smaller than the resultant page, stream, or conversation.
This changes, of course, with affordable HDTV cameras. That clip of the weekend at the beach is a massive file to move upstream, compared to typing in a web address and pressing “enter.” Peer-to-peer traffic also changes the scene. We’ll get to that.
The total available bandwidth of a contemporary cable system is likewise highly asymmetrical. It goes like this: Upstream traffic moves within a tiny gash of spectrum between 5 MHz and 42 MHz. Downstream stuff moves over a path that starts at 54 MHz, and goes as high as 1 GHz (or, 1,000 MHz).
Why 54 MHz? Why not just move that boundary up higher, to make the upstream path wider?
Here’s what veteran cable engineers say, when asked question: Because that’s where channel 2 starts.
The Regulations of Spectrum
What they mean is this: Long, long ago — as in 74 years ago, in 1934 — the Federal Communications Commission was empowered to develop, maintain, and enforce a table of radio frequency allocations for non-governmental use. (The National Telecommunications and Information Association handles the spectrum used by the government.)
This industry’s monthly engineering trades actually print a “Frequency Chart,” produced annually and in coordination with the FCC. Walk into any headend, and you’ll probably see one tacked to the wall somewhere. They’re a colorful bit of “tech art.”
As a direct result of the FCC’s frequency allocations, the U.S. cable industry is required to contain its upstream (or “reverse”) path traffic within that little gap, located between 5 and 42 MHz.
(The 12 MHz separating the top of the upstream band, at 42 MHz, and the bottom of broadcast channel 2, at 54 MHz, is called “guard band.” It prevents the two segments from colliding and making a mess.)
It’s true that cable’s spectrum is contained within shielded wires, and doesn’t free-wheel through the air, bumping willy-nilly into broadcast channels. Still, the FCC decided that cable operators should display off-air channel 2 at the identical 54 MHz spot as the broadcasters — so that ordinary people could find it, when they tuned in. (Remember — this was 60 years ago.)
So that’s the regulation part of why cable’s broadband plant isnt symmetrical.
The Physics of Spectrum
Let’s back up even farther — to the 1860s. That’s when a Scottish physicist named James Clerk Maxwell found a way to combine the properties of electricity with the properties of magnetism. His discovery was foundational to the “electromagnetic spectrum” that is the basis of all modern telecommunications. (To put this in people-context, Albert Einstein had two hero shots on his wall — Newton, and Maxwell.)
The electro-magnetic spectrum is invisible, and vast. It’s not just about radio waves. Microwave is there, and infrared. So are ultra-violet rays, x-rays, and gamma rays. In varying degrees, and if you could see them, they all look like the letter S, on its side. The sine wave.
The RF (radio frequency) portion of the electromagnetic spectrum can be made to do stuff — like carry radio and TV — by manipulating its frequency (the number of times the sideways S recurs) and its amplitude (how big the sideways S is.) This manipulation is called modulation — the imprinting of a signal, onto a wave.
In the cable upstream path, the type of modulation commonly used is deliberately and necessarily sturdier than what’s used in the downstream path. That tiny slash of upstream spectrum is a tough environment, bristling with noise and impairments. It’s the road with potholes big enough to swallow a Mini Cooper: You just have to slow down.
So that’s the physics of it.
Then there’s the empirical evidence, which says that people, on average, receive four times more information than they transmit. Maybe this will change, with user-generated video. Peer-to-peer traffic will, by its very nature, occupy all unused space. It behaves like a gas, that way.
The bottom line is this: Nobody is deliberately precluding symmetrical bandwidth. In the beginning, nobody even knew what to do with the upstream spectrum provided to them by the FCC. TVs were still black and white and analog. Two-way plant wouldn’t emerge for another 40 years.
But, you say: If the broadcasters are going all-digital in February, doesn’t the channel 2 thing go away, or at least present a way to expand the upstream boundaries?
Ah, wouldn’t that be grand. Alas: Going digital is one thing. Outfitting hundreds of thousands of amplifiers, taps, and in-home TV tuners to know that the upstream got wider is another.
This column originally appeared in the Platforms section of Multichannel News.