December 28, 2005

Om, Niall and Two Tier Internet

Om and Niall have posted a podcast where they express the thought that eventually the access providers will give preferential treatment to their traffic compared to others. Om suggests that while explicit degradation of others’ traffic will face political scrutiny, preferential treatment will not raise an eyebrow. He also indicates that this will happen within 12 to 18 months and cautions content providers to develop a strategy to handle the effects this will have on their service.

It is likely that access providers will indeed give preferential treatment to their traffic; but I am not sure about the predicted effect. Even though Quality of Service and prioritizing traffic have been talked about for a long time, the empirical data seems to suggest that congestion is not an issue in the core. So if the ISPs indeed give only preferential treatment (as opposed to blocking traffic from competitors) and a user is only accessing “open” data, then that user should not observe any change in the level of quality. Even those who access both sets of data will not see much difference as long as the amount of bandwidth consumed by the “protected” services do not exceed a reasonable fraction of the available bandwidth. This is not a big problem because it is strictly decided by each individual user and not by the ISP. In other words, preferential treatment of itself does not harm anyone. But that is not what the executives from ISP companies seem to suggest; they want to extract equal flesh from even those who are willing to abide by “best effort” service. Even though the advocates of Net Neutrality may not make it explicit, their demands are only in reference to best effort service.

I have a related thought. Net neutrality and best effort service are necessary and sufficient conditions for realizing the benefits of The Long Tail. If I am a producer in The Long Tail, the content I generate can be consumed using best effort service and my customers have unimpeded access to the net, then I am OK. Conversely, if either the application requires some special delivery service or special business relationship with the ISP is needed, then by definition The Long Tail is not viable. So if you are a believer of The Long Tail phenomenon, then you should demand Net Neutrality from the ISPs.

Posted by aswath at 05:05 PM | Comments (0)

December 12, 2005

Serverless Gets You

In a recent entry Andy tells us about a problem that I have suspected for a long time. Somebody left an “offline” message to Andy, but it took a couple of days for Andy to receive it. He suspects that the problem is started to happen recently. My opinion is that architectural and has been pointed out previously.

It is reasonable to conclude that a designated Skype client acts as an answering machine and collects the message. Then to ensure reliability may store copies of the message in multiple clients. This way, when Andy came on line, he could retrieve a copy. This may look like a reasonable alternative to a central voice mail server. Except there is a non-zero probability (for some reason engineers want to call this “finite” probability; isn’t probability finite anyway) that all the clients where a message is stored is not available when Andy signs on. (Think of a big holiday or weekend in Andy’s case). Actually, the first client may be disconnected from the network as the person is leaving the message without that person realizing it. It is all in the race conditions.

The fundamental problem is in trying to replace a server (that can be made reliable or at least whose failure is discernable) by a number of unreliable clients, however large the number may be. You see the law of large numbers is in play if these clients are independent and under certain conditions they may not be independent.

It is interesting that Andy has encountered this problem. Peerio, a system developed by one of his own clients may also have the same problem. Essentially they also have to do the same thing to earn the “serverless” moniker. I say this with a little trepidation: whenever I write about them, a representative from Popular Telephony comes swinging at me personally (read the comment by prostuda) without arguing the point.

A better alternative is for the client of the originator to collect the message and send it out as an email.

Posted by aswath at 09:21 AM | Comments (4)

December 09, 2005

Dodging Skype Draft

In a recent post, Om points out to a site (originally identified in Skype Journal) that plots Skype supernodes around the world. He also points out that these supernodes are a source of concern in many organizations and that some products and home cooked schemes have sprung up to address this. If there is conscription, people will find ways to avoid it. Some will learn to say “zed” and move north; but then some register for higher studies and avoid it. In other words, can one continue to use Skype but avoid becoming a supernode or a relay node?

For some time I was under the impression that if the client is behind a NAT/FW then it will not be a candidate. But in reality this is not the case. If the device is behind a well “behaved” NAT, then it effectively has a public IP address/port and so is reachable by others. This means that such devices can be used as supernodes. So the only effective way is to place the Skype client behind a symmetric NAT. One of the questions Skype Journal asks is “Why are Chinese Skype supernodes not shown on the map? Where else are these factors at play?” I think I know the answer. I am told (and I have no way to verify) that Chinese ISPs deploy their own symmetric NATs. This probably explains the absence of supernodes in China.

There is another interesting observation. Om refers to a mailing list post that describes a simple and open way that an enterprise can block Skype. Given this, I wonder what is so special about Verso, which got written up in Wall Street Journal among others.

Posted by aswath at 11:15 PM | Comments (0)

December 07, 2005

FWD Could be a Platform for Voice 2.0

Less than three weeks back, I wrote about beta testing of Relevance Engine from iotum. At that time I mentioned that iotum need not restrict the beta to only those who reside in one of the few area codes if only they partner with a VoIP provider like FWD. Past Monday, iotum and FWD announced just that. They are scheduled to conduct the trial in January. So it is reasonable to assume that they have been working on this for a while now. In two recent posts, I had suggested other services like SlingBox and FolderShare can partner with FWD. Probably the partnership with iotum could be used a model for others as well. There is no risk of undermining the integrity of the whole system because of these trials. After all, trial participants can be isolated to a separate proxy. This way, FWD could be the laboratory where more and more Voice 2.0 (actually the label is myopic) services and features can be experimented, thereby earning the moniker “for geeks by geeks”.

Posted by aswath at 11:41 PM | Comments (1)

December 05, 2005

Nope to Skype?

Russell Shaw points to a story in that claims that “Big eBay users shun Skype”. As usual, if we dig just a bit deeper, we realize that the truth is a bit less categorical than that.

The original article quotes the executive director of a group of high-volume eBay sellers as saying that "Skype doesn't give me a capability that I already don't have." Apparently this based on informal conversation with members. Of course it is not clear why they should feel this way. Is it because they feel that Skype calls will cost them more in commission? Is it because they sell the garden variety of items and such calls are not needed? If I am occasional seller and have not earned enough trust, probably that is when I will benefit from Skype, because the potential buyer will overcome the initial hesitation after talking to me.

In other words, this story doesn’t give enough supporting evidence for their claim. Nonetheless, I feel that eBay sellers can use third party VoIP tools as well instead of Skype. So I still feel Skype is not going to change much the landscape for eBay.

Posted by aswath at 06:09 PM | Comments (3)

What is IP Communications?

On the heels of writing my previous entry where I claimed SlingBox is an IP Communications device, I read a story (registration required) in New York Times that discussed different tools that allow users to synchronize files between multiple computers. One of the applications they talk about is FolderShare. Om Malik and Alec Saunders have talked about it sometime back. In my opinion, this is another instance of IP Communications. Jeff Pulver has talked about it many times. But I don’t recall whether he has given a definition for this term. This entry is an attempt to capture the specific characteristics of IP Communications.

First a quick synopsis of FolderShare (as usual, I have not used it; it is based on the description available in their website): You establish an account with FolderShare and install their client in all the computers of interest. Then you identify the files that you would like to share among the computers. From then on these files will be synchronized between the computers. The synchronization works within a LAN or across the Internet. They use UPnP and port forwarding mechanisms for NAT/FW traversal. They even claim that it uses P2P technology. Originally there was a membership fee for enhanced service; but after the recent acquisition by Microsoft, the service is free.

SlingBox, FolderShare and countless other services seem to have the same pattern: basically they all transfer data from one device to another. Yes the specific application is different; they may differ in the performance requirement. But essentially they are all data transfer engines. The second common aspect is that they all have some sort of NAT/FW traversal technique. Finally they have some sort of authentication and authorization mechanism. This set of observations suggests the general characteristics of IP Communications. IP Communications is one where two are more end-points exchange data having the following properties:

  1. An identification mechanism which can be used to authenticate users and also can be used to authorize the set of other users who can be contacted. The authorization rule can be static or dynamic (Relevance Engine anyone?); it can be application specific.
  2. An efficient way to overcome the NAT/FW boundaries without violating the security policies in place.
  3. A set of application specific data generator that is used to encode/decode the data.

There is a benefit in developing this general definition. To date most of the examples of IP Communications have ad-hoc or proprietary way of realizing the first two functions. One consequence of this is the users are required to keep many ids. It is worse that the application providers use this to create an artificial “walled garden” in the hope that this will give them some economical benefit. Instead, it is preferable that the industry uses standard mechanism for these functions. For example, we could use SIP methods for authentication, authorization and the related directory services and we could use ICS for NAT/FW traversal.

Just imagine the possibilities. Currently many VoIP systems like FWD and SIPPhone offer the first two functions and they allow for federation with other systems. It is true that they offer “data generators” for voice, video, IM and Presence. But they could easily allow third parties like SlingBox and FolderShare to be “helper applications” that use their clients. There will not be appreciable increase in the demand for FWD/SIPPhone resources. But the user will experience the true benefits of IP Communications.

Of course, I used FWD/SIPPhone as examples. Instead I could have used Yahoo Messenger or GoogleTalk or MSN Messenger. I guess that explains why MSN recently acquired FolderShare.

Posted by aswath at 05:54 PM | Comments (1)

December 04, 2005

Graduating from 9 out of 9 to Applause

Andy Abramson is happy to report that PhoneGnome won an award in the multi-media gadget category in Gadgetfest 2005. As Andy tells it, it tied with SlingBox during the first round of voting and the tie continued for two additional round of voting and then won on the third one. I feel if we analyze the features and user experience of SlingBox, we will see how it could have continued the voting process for additional rounds and end-up in a real tie.

Drew Terry gives us some additional details: There were five contestants. Three of them are of interest for this discussion. Apart from PhoneGnome and SlingBox, the third entry of note is Sony’s Location Free TV Solution. Interestingly, both Slingbox and LF offer the same capability – place shifting media. Since I do not have hands-on experience of any of these three products and my knowldgebase is only from the respective user guides, let us assume that all the three products perform the central functions admirably. But there is a world of difference in the user experience while connecting the box for the first time.

Connecting PG is the simplest. One just needs to connect the power cord, telephone line and Ethernet cord. Everything else transparent to the user and takes place in the background. Its id is determined by the number of the associated PSTN line. SlingBox is slightly more complicated. The user needs to know what is UPnP, how to enable it in the router and setup port forwarding in the router. It is id is an unbelievable 32 characters long, assigned by the factory. LF is the worst, at least as it is written in the user guide. One has to worry whether the place shifting is happening within the LAN or over the Internet. One even has to worry about Dynamic DNS! Though there has not been much written about LF, Om Malik and Jeff Pulver have written about their experiences. Even though they are generally positive about SlingBox’s video quality, if you read the comments to Om’s post, you will realize network setup could be simplified. Even Jeff had to struggle with port forwarding. Can you imagine his reaction if he had to do all this for a VoIP device?

If you really think for a minute, there is no fundamental difference between PG, SlingBox and LF. They all need to provide directory service and have to device a mechanism for NAT/FW traversal. PG uses SIP for the former and most likely uses Session Border Controller for the latter. It is immediately clear that SlingBox and LF can also use SIP for directory service. Of course, it is not optimal for them to use SBC. After all there are times when SBCs have to handle the media flow. This is not preferable since video will be bandwidth intensive. Instead it is better that they use ICE. ICE will address the LAN issue and as long as both the SlingBox/LF and the client are not behind different symmetric NATs, it is a scalable proposition to host STUN/TURN and assist in NAT/FW traversal. If SlingBox incorporates these suggestions, then it will surely tie in GadgetFest 2006, if not get the award outright.

Posted by aswath at 01:48 AM | Comments (0)

December 01, 2005

Misleading Headlines

In print journalism, it is not uncommon for headlines to prejudice the readers and convey an erroneous point of the story. Internet being a fast world, this happens routinely there. For example, it is indelible in the minds of VoIPers that VoIP is illegal in India. No amount of explanation will be enough. This morning Om Malik has fallen victim to this in a single post – twice actually.

In the post in question, Om discusses Skype 2.0. In the passing, he states, quoting Stuart Henshall, that some Skype customers’ passwords have been compromised. If you go to the source, Stuart concludes this based on an advisory issued by Skype Security System. If you read the email, they do not claim this at all and it much less dramatic. A sister site has not followed the security guidelines and so they have taken an extreme procedure. There is no evidence that some third party got hold of Skype logins and passwords. Their corrective procedure may have inconvenienced some users; but that is another story. The real story here is that Skype is the owner of your identity and it can keep you out of their world at anytime for any reason. It is not P2P in that sense – it is P2P only when it is beneficial to them. But this point may be esoteric and may not be so catchy.

The other item is related to the news item regarding Israeli Communications Industry directive requiring all ISPs to “ensure that their systems are not illegally supporting long-distance calls using voice over Internet (VoIP) technology.” Both Jeff Pulver and Tom Keating have discussed this story and have concluded that Israel blocks VoIP. A Ministry representative has to individually contact them to inform them that what is prohibited is illegal interconnection to PSTN, not even all forms of interconnection. But Om’s source goes one step further. Their headline states, “Ministry out to throttle free Skype calls”. In the body of the text, Skype is mentioned just once when they say that “[VoIP] allows companies like Skype to offer free international phone calls over the Net.” My request to fellow VoIPers is that if we really believe in the revolutionary nature of the technology and if we think rate arbitration is not really the point of VoIP, then let us ignore PSTN totally. Then any restriction on interconnection to PSTN will be a non-event.

On the other hand, we should pay attention to blocking open communication in the IP domain as reported by another story Om refers to in his post. It seems there is a company called Bitek International that can block Skype or any other designated application.

Posted by aswath at 08:44 AM | Comments (0)

Copyright © 2003-2009 Moca Educational Products.