Continuing on the theme set by Jeff and followed by my post I came up the following only half in jest. No disrespect is meant to Dr. King for I have enormous respect for him. I hesitated to post it yesterday. Initial reactions from some of my friends were positive. So I decided to go ahead. I hope that you also don't take offense.
Two days later: Originally I adapted Dr. King's "I have a dream" speech to suggest that in a packet switching world, packets could be differentiated only based on the network resources consumed and not specific applications. But decided to delete it after Bob expressed his comments. I have really made my point in my previous post.
Yesterday Paul Kapustka reported that wireless broadband provider ClearWire may be blocking Vonage calls placed by their subscribers. Referring to this story, Jeff expresses his concern that “Net Freedom” may not be available for customers of non-Title 2 service providers. He also recalls a couple of service providers telling him that “allowing VoIP service (free riders) which wasn’t associated with their own service wasn’t in their own best interests” and wonders how to solve this.
Jeff knows what to do, because he has done it before. Not too long ago many regulators and lawmakers used the “duck test” to claim that VoIP must be regulated and taxed. Over a period of time, it was pointed out to them that in IP network, voice communication can take place in different formats, like games, that are not regulated or taxed. Also it took some effort to establish that it will be difficult to distinguish regulated and unregulated traffic. I suspect that the situation is analogues here. It looks like since VoIP has voice in it, many ISPs seem to think VoIP is a source of revenue. ISPs need to realize that as long as they allow one kind of traffic, like web traffic, then it is possible to exchange VoIP traffic in an analogous manner. It is futile to distinguish and charge differently when the network offers the same level of service. It is better to charge on the total bandwidth consumed. By the way some twenty years back, Bells charged local voice calls over ISDN, just because they wanted to charge data calls (those days 64 kbps was “broadband”), even though both the calls consumed the same network resources. If only the two calls had been more rationally charged, I suspect that ISDN would have been more successful.
In the VoIP space, “developing country” is a put down phrase if a country regulates VoIP service providers in any shape or form. Panama, India and lately Costa Rica (“otherwise an oasis”) have been on the receiving end of this condemnation. Now comes a story that FCC may reject the forbearance application from Level 3 regarding access charges. Depending on how the order is written, it seems it could be disaster because the incumbents could sue many of the VoIP operators for retroactive charges. (This is a bit confusing because an FCC official, Robert Pepper said in VON that one of the myths on the industry is that VoIP service providers do not pay regulatory fees. Probably I misunderstood him.) It will be interesting to see how SkypeOut rates will be affected, because their rates are low not because of the technology but because of the regulatory arbitrage game played by the wholesalers.
James Seng has posted a sweet story (friendly advice - don’t visit the original link) of an experience of a kid in a Japanese school. It seems the only acceptable answer to the question “Why does the shadow of a stick in the ground moves over the course of a day?” is “the sun moves in the sky” and not “earth is moving”. In narrating this incident, the parent is justifiably happy that his son has learnt “an important lesson about not automatically accepting what he hears.” I hope that many of us in the VoIP arena follow this maxim and challenge the VoIP orthodoxy. In this spirit, let us look at an interview given by the CEO of Skype (via Andy).
In answering a question about voice quality that Skype will enjoy when everyone has moved to VoIP, Zennstrom says that unlike some VoIP systems, Skype traffic does not have to go “through a server to connect to each other.” He states that the users are connected directly and take “the shortest path over the Internet”. But we know this is not true, especially if the users are behind symmetric Firewall/NAT. We are told that one of the fastest growths of broadband is in China. I am also told (I have no way of verifying this) these users are behind a symmetrical Firewall/NAT at the service provider. If this information is correct then Skype has to employ supernode for every session and these supernodes must be from a smaller population. If majority of Skype users keep their sessions up all the time (as James and Martin suggest that they do), I pity the Skype users who have been drafted as supernodes. (Skype does not do optimization based on voice activity detection.) Of course one can avoid being drafted by artificially placing the client behind a Firewall/NAT; but then the viability of service is threatened.
Interestingly, Zennstrom revealed in the interview that Skype is working on a router embedded with Skype application. This is a smart move because these Skype embedded devices do not have the problems associated with Firewall/NATs and it is easy to enlist them for supernode functions, even if they have dynamically assigned IP addresses.
In response to a question about the impact regulation will have on Skype, as a true member of VON coalition, Zennstrom explains that there is no reason for regulation. Well and good. But then he adds that the “well-developed regions of the world” (I guess that now there is a gradation in the “developed” world) has this understanding, but the story is altogether different in developing countries. It seems that some of them “view telephony as a way to make money”, but he reassures us that “this is short-term thinking”. This is from the CEO of company whose revenue stream comes only from terminating calls to/from this telephony network (keep in mind that SkypeIn is a pure revenue source because of the regulations of the “brain dead” PSTN). I would like to know which developing country has taken a regulatory locus that is different than the one taken by FCC in US. (Please don’t give Panama as an example, because during initial days, US also discussed about regulating VoIP; only after the infeasibility was demonstrated, the stance changed.) If we can not name such a country, then shouldn’t we stop talking this us and them nonsense?
By the way it is official that Skype uses wideband codec, though the specific codec has not been revealed.
Rich Tehrani informs us of three new services Packet8 has added to its hosted PBX (otherwise known as Centrex) service.
The first service, Virtual Extension is an extension number that accepts only incoming calls and calls to these extensions can be forwarded to any other number by incurring per minute charges. Each extension is $9.95 per month. (Remember, that the service provider receives access charges for terminating calls, meaning each Virtual extension is a pure revenue source.)
The second feature, Metered extensions incur a lower monthly charge ($19.95 instead of $39.95) but receive a limited (250) minutes, with $0.039 per minute beyond that.
The third feature is porting of toll free numbers from an existing provider.
Rich’s concluding remark is noteworthy: “My take is, FINALLY! Let’s stop focusing on price. It’s all about the services… [t]he ones you can’t get over the PSTN. Sell the services first and tout the price last. That is the sure way to make it as a VoIP service provider.” Is he suggesting that one can not get these features in PSTN? Centrex or PBXs can’t really offer these features? Is VoIP label going to lull us from even a basic level of critical analysis? I say that the search for purple minutes hasn’t yielded much result.
Two weeks back, Robert X. Cringely wrote a column discussing the predatory actions that access providers can take (have started to take?) against upstart VoIP providers. Since Cringely is an iconic figure in the technology industry and this piece appeared under the title of “The Pulpit”, it will be useful to critically review his analysis. As is true of any self important and soapbox lectures, this one contains statements that make you laugh and some that make you cry – mostly at the reasoning used by the speaker.
First let me summarize the essay for ease of reference. Many upstart VoIP providers, called “parasites” (he attributes the label to the incumbents, when it is really the “high priest” of the industry, Pulver, who coined this term and implored the incumbents to become one; but why clutter our minds with facts) are undermining the incumbents’ business. In retaliation, the incumbents are offering their own VoIP services and are handicapping their service by giving preferential treatment to their traffic. Unlike port blocking, this is not considered illegal and results in poor quality for the customers of the parasites. The only recourse the parasites have is to quickly develop innovative services that the incumbents will be loath to develop.
He seems to imply the major benefit of VoIP technology is low cost long distance calls. Even he admits this low cost is mainly facilitated not by technology, but by favorable regulatory environment. What a shame. IP Communications can make so many features available to its users. Why doesn’t he point out these features and lament about the lack of activity in this regard?
When AT&T launched its CallVantage service, their press releases never ceased to point out that unlike other parasites, they carry the VoIP traffic over their private QoS enabled network. So I suppose that there is a gradation even among parasites and that providers like AT&T will not be affected by the antics of the incumbents.
He states that this “web services war” will in the future be waged against services related to movies and music, impacting CinemaNow and Bit Torrent. He does not elaborate how lack of QoS will impact “store and play” applications. Of course he doesn’t explore the incongruence of QoS enabled network and “dumb” IP network. There is no clue as to what happens to a QoS enabled network when all applications start to demand higher levels of QoS. Is there any difference to a circuit switched network then? We can’t turn to his sage counsel.
He gives an example of an innovative service that the parasites can develop to counter the onslaught of the incumbents. It is a service from theswitchboard.ca, where two users can download an applet in real time and use it to “listen to each other’s heavy breathing” (what a contribution to the social welfare) or whatever. This way there is no need to download and install an application a priori. Of course he does not suggest how this new innovative service overcomes the predatory action of the incumbents that overpower parasites’ VoIP service offering.
The problem with this analysis and other similar analysis is that it continues the “bellheaded” thinking that there is gold in POTS. With IP Communications there is no additional money in voice communications. If we reconcile to this, then as a group we will not be distracted by the gold rush and would be developing features that truly revolutionizes end-to-end communication.
Andy writes that the million or so SkypeOut users get “less than a great experience with sound despite GIPS being in place.” It was known long time back that SkypeOut uses G.729A and that its quality will not be as good as Skype calls because the latter uses a wideband codec. Looks like the interconnect providers still do not see the need to upgrade their gateways to support the wideband codec.
While we are on the topic of codecs, one of the issues with ATA based deployments is that it is not feasible to add wideband codec to the ATA, since the standard telephones sample only at the baseband. I hope that the new breed of cordless phone based IP phones pay attention to this and build their handsets so that they can operate at the wideband frequencies.
Andy points to a press release from Verisign and Vonage that announces that Vonage has signed up Verisign’s NetDiscovery service. If you already do not know about NetDiscovery, it simplifies the backend operations – interfacing to the Law Enforcement Authorities (LEAs) to receive subpoenas, delivering intercepted information to them and the like. These are all very complicated and there are many rules and regulations. In that respect, this service from Verisign is very advantageous to service providers like Vonage. But that is only half the story.
The other half is the effort involved in collecting the call control information and call content information. Even in here, Vonage will not have any difficulty in collecting the call control information, because they process this as part of a normal call setup. But collecting the call content information is another matter altogether. If the targeted call is transitioning to PSTN, then the Gateway is a natural intercept point. But if the Gateway is provided by a third party, then Vonage has to introduce the intercept point. If the call is end-to-end IP the problem increases many fold. In this case, there is no natural intercept point. This means one intercept point has to be introduced. Since one of the CALEA requirements is that interception should be imperceptible and the user can easily detect that an intercept point has been introduced, the service provider must use the intercept point for all calls – whether they are targeted or not. It is not just the capex has increased for the service provider; the media traffic has to be brought into the service provider’s network and then carried out. In other words, Vonage can not claim, they do not have any network infrastructure.
The other important implication is that with the introduction of the intercept point for all calls and taken together with the SIP Proxy, VoIP service provider has recreated Class 5 switch and has truly and irrefutably gone back to the “intelligent network”.
This analysis applies only to non-facilities based service providers. For DSL and cable modem providers, DSLAM and CMTS are natural intercept points, without impacting the capex and opex expenditures. So if the CALEA requirement is extended to end-to-end IP calls as well, facilities based providers will have an advantage. This issue must be taken into account by the lawmakers and an appropriate solution must be developed. One possibility is for the LEA to coordinate the services of VOIP service provider and the access provider.
Popular Telephony has developed a serverless computer/communication technology called Peerio. They have talked up the benefits of this technology without giving much detail, either on the specifics of the technology or on the specific advantages. The first application of this technology seems to PBX replacement for SMBs. Many have expressed doubts about Peerio because the previewed software has many implementation problems. Previously I had questioned the need for a serverless communication system. Recently I summarized some of the technology behind Peerio as can be surmised from one of their patent application. A commenter (“prostuda”) to the post suggested that I missed the most interesting points, “redundancy on top of serverless network” being one of them. Subsequently, Martin wrote that “the serverless concept is a red herring; the important thing is self-configuration”. Nonetheless, he also thinks real-time communication is the major application of Peerio.
As serendipity would have it, C|Net recently wrote a story about the technology Google uses to manage failures at their data centers. According to this article, Google has networked multiple standard grade computers and data is replicated in multiple places. Is this what “prostuda” and Martin have in mind? Is the real point of Peerio technology to replicate Google technology for storing and retrieving data in an enterprise? If that is so, won’t they be focusing on large enterprises, rather than SMBs? Probably some their planned announcements in VON will shed some light on this.
According to Reuters (via eurovoip), Skype users get free wifi access from Broadreach to make Skype calls. It seems normally Broadreach charges 3 pounds per hour. I wonder whether there is an arbitrage opportunity here if one can tunnel an IP connection through a Skype connection. Consider a potential wifi user sets up a “fake” Skype session to a proxy which can access Internet. This is possible if they allow a text chat session as well.
It seems that AT&T CallVantage has added directory service. They must have taken a press release (couldn’t locate it on their website), since both Andy and Rich have mentioned it in their respective blogs. (Not to be outdone, Packet 8 announced they offer the same service for 75 cents; Vonage (probably many others) offers it for who knows how long). I am sure the brain trust in AT&T will soon enhance this feature so that a call will be established based on the result of the query without the user dialing the numbers.
Is this what we can expect from the revolutionary technology of VoIP and the Stupid Network, Internet? Since VoIP device is always connected to a network to which Google is also connected, can’t the device query Google and get the directory entry for free? It seems that not just the network has become stupid and the intelligence at the end device has been squelched, but also by waving the VoIP flag, the users can be mesmerized into stupidity.