Quick & Swift Messengers are no godlike beings like the god Mercuy or the all powerful Hanuman. Cable networks, data distribution centres and many more things need to function perfectly to make the being here & there at the same time possible.
For those who took ubiquitous connectivity for granted, the land of failure is always in sight, right over the horizon and today we have a glimpse of a world without SKYPE…
Ardent lovers will not be affected they will switch over – with ease – to telepathy.
Note made a few hours later when SKYPE has been going on for a few seconds and off for many minutes several times and skimming the surface of SKYPotology and its technical feats & shortcomings, I bump into this dated article, taht still seems to point to a way to come to an insight. Let me quote and pass on the link + the interesting map all on ‘Accelerating The Internet‘ it is from 2007, but it may help to link to newer researches by these knowledgable people:
Peer-to-peer (P2P) applications are consuming a significant fraction of the total bandwidth of Internet service providers (ISPs). The increasing P2P traffic is becoming a financial burden to the ISPs and if not well addressed may lead ISPs to block or put strict rate limits on P2P traffic.
An Interesting Idea: PCP Framework
PCP: ISP-compliant, Peer-friendly P2P
We address this important issue by proposing a new framework, PCP, for designing P2P applications that smoothly fit into the global Internet. In the framework, an ISP decides on how much of its bandwidth is to be allocated to P2P applications, and P2P nodes inside the network query the allocated bandwidth and adopt a peer-friendly algorithm to fairly share the allocated bandwidth.
How PCP can help both ISPs and P2P?
Using the typical percentile-based charging model widely used in the Internet and real traffic traces, we show that an ISP can allocate a large amount of bandwidth to P2P traffic without increasing its financial cost. We also show that a distributed, peer-friendly algorithm can be implemented to share the allocated P2P bandwidth.
What is wrong ?
It has been increasingly popular to build voiceoverIP (VoIP) applications based on peertopeer (P2P) networks in the Internet. However, many such VoIP applications freeride the network bandwidth of Internet Service Providers (ISPs). Thus their success may come at a cost to ISPs, especially those on the edge of the Internet. Their performance may degrade if ISPs take proactive measures to prevent freeriding.
Using largescale end-to-end measurements, we first conduct a systematic analysis of Skype supernode network. We then investigate the impacts of the access capacity constraint and the AS policy constraint on the VoIP quality of Skype. We show that even when freeriding is no longer possible for only 20% of supernodes that are located in stub ISPs, the overall VoIP quality of Skype degrades significantly, and a large percentage of VoIP sessions will have unacceptable quality. This result clearly demonstrates the potential danger of building VoIP applications based on P2P networks without taking into account operational models of the Internet.
How to make it right?
Time diversity in traffic patterns can be used to be (implicitly) ISP-compliant. We show that when using time diversity in traffic patterns in choosing supernodes, about 75% of supernodes in stub ASes can still be active after using time diversity in the experiment; as a result, about 15% of previously inactive supernodes become active, leading to approximately 60% increase in the relay capacity, which results in the observed VoIP quality improvement.
Another source that helped me a bit to understand what might be the problem with SKYPE is “A Brief Primer on the Tech Behind Skype, P2PSIP and P2P Networks” dating from November 2010. We may try to understand at least what the technology is we are using and at the same time make some sense of the fights over new regulations of the Internet in the USA that also touch on technologies described here…