P2P Part 2: Good, Bad On Byte Caps - Multichannel

P2P Part 2: Good, Bad On Byte Caps

Publish date:

Say you're in the business of moving electronic things from one place to another. Like from a headend to a home, or vice versa. (This should sound familiar.)

One day, you're told that your transit pipelines are filling with silt — at an alarming rate.

As tools go, you probably need either a good mesh, or a roto-rooter.

Welcome back to the sticky business of what to do about peer-to-peer (P2P) traffic. Last time, we examined why and how this P2P phenomenon is happening. This time, we'll examine what can be done about it.


P2P, which happens when people apportion parts of their PCs to share files with other PCs, is clogging up cable's broadband pipes in a rather big way.

How big? Chances are high that right now, as you read this, 50% of the traffic moving through broadband networks (digital subscriber line is not exempt) is P2P. That's especially true in the upstream signal path, which is cable's scarcest bandwidth resource.

The good news: There are now two things you can do about it. The better news: An increasing number of suppliers are developing "data forensic" tools, which delve deeper into the situation.

The first option is to implement a "byte cap." (Sadly, this is not something you put on your head.) Rather, it's a counting mechanism: Bytes received and bytes transmitted, per customer.

The field varies on byte caps. Cox Communications Inc.'s limit is a generously-sized 30 Gigabytes down and 7.5 Gb up, per month. To put that in context, your laptop's hard drive is probably in the 30-Gb range.

If you're going to institute byte caps, technologists caution, be sure to begin by putting counters on individual cable modems and on the spigots of the CMTS (cable-modem termination system) at the headend. Counting packets is relatively new. Miscounting: not good.

The second option is a direct outgrowth of byte caps: Upselling customers to a higher data tier. This is the stuff of the DOCSIS (Data Over Cable Service Interface Specification) 1.1 cable-modem upgrade, which tiptoes ever closer to the marketplace.


What happens when a broadband Internet customer exceeds the maximum number of allowed bytes per month? Now, not much. The byte cap and tiering phenomena are still fairly young. But the thinking is, on first offense, customers get a warning: You're over. Want to upgrade?

(If the lawsuits doled out by the Recording Industry Association of America three weeks ago are any indication, it is precisely at this point that many broadband grandparents start to get wise about what their grandchildren are doing on Grampa's PC.)

On the second overage, maybe customers get a harsher warning: You're way over. Upgrade or else.

Maybe the third time is a charm (ahem) not unlike your cell-phone bill, after you exceed the allocated minutes: Just as you never realized how expensive extra minutes could be, broadband customers realize that you're not kidding about bandwidth overages.

Part of the trickiness about P2P is how to notify customers about overages. In some cases, customers are honest-to-God, no-kidding, completely unaware of their "offenses."

As with data forensics, there is vendor help out there for customer messaging. Texas-based PerfTech is one, which accomplishes its work without using e-mail, instant messaging, or telephone contact. It acts sort of like a browser redirector, which an operator can customize to say whatever it needs to say.

Brute force

Tiering and byte caps are viewed in the data community as "brute force mechanisms."

That's because a big part of the problem is that P2P bits often look exactly like Web-browsing bits. Using the pipeline analogy, it all looks like water, even though half of it behaves like silt.

And, because P2P bits don't crust on the walls of the big pipes — they keep on moving — the remedy isn't like angioplasty. What's needed is more like a magic mesh: Something that sees color in clear; patterns in uniformity.

Again, the vendor community is responding. Companies like Ellacoya Networks, P-Cube, and Sandvine are all building CSI-like tools for P2P and broadband.

That's good, because that P2P traffic percentage – half the pipe — is huge. Even the engineering community, a predictably unflappable bunch, is alarmed. "It completely blew me away," said one MSO data technologist recently, when he learned that half his traffic is P2P.

Maybe it's just me, but, when I hear the seriously smart and sensible of our industry's technologists say things like "it completely blew me away," I listen more closely. In case you've never tried it, know that it not easy to "completely blow away" an engineer.

If the adage is true — that "anybody can do the job with the wrong tool" — then it's probably time to start gathering the right tools for P2P.

Stumped by gibberish? Send translatables toEllis299@aol.com.