Value Your Bit Pipes

We have all heard it said that telecom operators must move up the value chain if they are going to increase their corporate revenues and compensate for declining or stagnating voice-centric sales. I have heard or read statements from dozens of senior service provider and equipment vendor professionals in recent years highlighting the need to add value beyond the commoditized bit pipe. Former Vodafone Group plc (NYSE: VOD) CEO Arun Sarin captured this sentiment when he reportedly told an audience at the Mobile World Congress earlier this year: "We must not allow ourselves to become bit pipes and let somebody else do the services work."

If you think about being a dumb bit pipe only for voice, that warning makes a lot of sense. But in today's interconnected, data-dominated world, there is more value in the network than ever before, because of the wide variety of mission-critical and other services and applications that can be delivered over bit pipes. Instead of surrendering the high ground in public forums to companies such as Google (Nasdaq: GOOG) and Yahoo Inc. (Nasdaq: YHOO), facilities-based service provider and cable executives should be increasingly emphasizing the enormous value associated with delivering bigger, smarter, and more reliable bit pipes to consumers and businesses.

Network operators need to recognize that they are collectively sitting on fiber-based oil fields, and the big pipes they provide carry the digital fuel that drives information-based economies. Individual value-added services and applications will come and go, but the fiber networks and their optical/Ethernet bit pipes will not disappear for a long, long time.

We're talking about a paradigm shift in thinking here that requires telecom operators to be willing to consider what their networks are worth in light of macro-economic trends that place them in the driver's seat. They need to stop whining about the applications guys riding on top of their networks for free and, instead, methodically set out to reshape the public debate.

Banging the drums even louder for a national broadband policy, for example, would be one way to highlight the value of network connectivity and the critically important role of big bit pipes in driving productivity.

It also makes sense to raise consumer broadband prices at least at the rate of annual GDP growth. There is no reason why telcos should not recapture a small portion of GDP growth associated with broadband-enabled efficiencies in economic transactions.

— Stan Hubbard, Senior Analyst, Heavy Reading

Page 1 / 2   >   >>
krbabu 12/5/2012 | 3:35:30 PM
re: Value Your Bit Pipes Stan:
You are right on the money.
Historically, the operators have always focused on providing only pipes. It is time that they focus on adding greater value; I don't think they need fear of alienating their customers.
elisabetwahlgren 12/5/2012 | 3:34:51 PM
re: Value Your Bit Pipes I suspect a great number of customers would be willing to pay (a lot) more for guaranteed QoS. Delivering quality real-time services is what telco operators used to be both knowledgeable and passionate about, and I sincerely wish they would make use of this competence, if it is still there (if they haven't already kicked out all experienced staff).
rogerdavis 12/5/2012 | 3:34:50 PM
re: Value Your Bit Pipes The bit pipes we have should be like the motorway / road network, charging for faster lanes, shorter routes, but giving a public service for a reasonable fee / tax .
with Deep packet inspection devices this can be achieved so those who require higher spec services can run them at correct pricing and we should all get win/win
mgardner750 12/5/2012 | 3:34:48 PM
re: Value Your Bit Pipes I couldn't agree more on everything you said in your article except your last paragraph.

The network is an extremely valuable asset, but the way they are trying to extract value from it is sub-optimal.

fgoldstein 12/5/2012 | 3:34:47 PM
re: Value Your Bit Pipes Roger,

It is perfectly reasonable to offer bit pipes of different types, with different prices. This is the traditional Telecom business model, as embodied in ATM, which had multiple QoS options, cell rates, etc. However, the Internet came along with its all-you-can-eat business model, and it took enough of the volume (the preponderance of bits) and thus killed interest in the non-voice aspects of Telecom.

DPI is entirely, completely, and utterly the wrong way to go about this. The post office does not open our mail, read it, and decide how much to charge or how fast to deliver it! But they do offer multiple services. (And lest anyone argue, yes, certain classes are priced based on the percentage of advertising, which the mailer states.) DPI is simply wiretapping, and, absent a court order (such as CALEA), should be treated as a crime. If there were a network offering of distinguished QoS (i.e., something other than the Internet's current worst-effort model), then it should be signaled explicitly, via a connection request tied to the packet header.
Mark Seery 12/5/2012 | 3:34:45 PM
re: Value Your Bit Pipes There appears to be a tilt towards suggesting all uses of DPI are an absolute evil. There are many things that could be broadly viewed as DPI: ACLs, Security, SBCs,....the "envelope" for a service does not necessarily end at the 5 tuple.

If an operator can add value to my service by using "DPI" and I have given permission for that, then why would anyone want to outlaw that unless they were pursuing another agenda - like structrual separation via technology fiat for example?

This processing will be done somewhere in the network, by banning facilities operators from doing it we are just picking winners and loosers in the market, and moving the regulatory problem somewhere else, and if so that should be openly discussed rather than obfuscating it.

The focus should be the behaviour / policy, not the technology. Policy/principles are more portable across technology innovations and broadly applicable across food chain.
rjmcmahon 12/5/2012 | 3:34:44 PM
re: Value Your Bit Pipes re: "The focus should be the behaviour / policy, not the technology."

Agreed. But just because there is money signal on the other end of behavior doesn't mean it's appropriate behavior needed for progress. The faith in market as God instantiated doesn't work either. Begs the question though, doesn't it?
rjmcmahon 12/5/2012 | 3:34:44 PM
re: Value Your Bit Pipes re: "(i.e., something other than the Internet's current worst-effort model)"

The irony of course is that today the worse-effort model beats the engineer's pipe dream by a multiple of 240 with respect to performance and at least an order of magnitude w/respect to cost.
Mark Seery 12/5/2012 | 3:34:43 PM
re: Value Your Bit Pipes "SBCs, ACLs, etc., are perfectly normal functions."

Thanks Fred, glad we are in agreement about that. The other fgoldstein had me a little worried ;-)

"There is never a need to do Deep Packet Inspection (that is, above the TCP/UDP layer). Application bits should be sacred, but the nature of the TCP flows can be monitored."


Not sure we have the same view of how deep in to a packet headers can be found or the finer points of TCP/IP layering, but I don't think they are really worth arguing about in the context of this discussion...
fgoldstein 12/5/2012 | 3:34:43 PM
re: Value Your Bit Pipes Mark,

> There are many things that could be broadly viewed as DPI: ACLs, Security, SBCs,....the "envelope" for a service does not necessarily end at the 5 tuple.

If you take something normal and useful, and define it as belonging to a category that it doesn't belong in, then that category suddenly has a normal and useful purpose. But then again if you call a leg a tail, a dog still only has four legs, since it doesn't really matter what you call it.

The term deep packet inspection was coined because it went beyond the headers, deep into the payload. If it's done in headers, it's not DPI. It may be shallow packet inspection, stateful routing, or all sorts of other things, but you shouldn't call them DPI. I don't object to doing fun stuff in the headers; that's what they're for. Orthodox neuts who are stuck in 1983, like FCC Freak Show star Dave Reed, probably think it's not okay to even look at TCP headers, but that's really not a layer violation; the TCP/IP split is one of implementation, not a proper layer boundary.

DPI itself is intended to do two things. One is replace network-layer relaying with application-layer relaying, so that the network can discriminate between applications regardless of their TCP port. For example, a wireless carrier can charge for SMTP and POP (to the customer's own server) based on the messages sent, not the volume, since they compete with SMS. That is bad enough.

The other, also pitched by DPI vendors (have you seen them pitch?) and requested by the likes of AT&T, is to extract "value" from communications by extracting information from the payload, beyond the header, that indicates the nature of the flow. For example, take a cut out of an online purchase or bank transaction.

Those are both wiretapping and should be treated as criminal activities. SBCs, ACLs, etc., are perfectly normal functions.
Page 1 / 2   >   >>
Sign In