DisUnified Collaboration

The Internet Wins because it is a decentralized system of redundancy that is delivering a highly available platform. We could make a case that Peer to Peer vs IP based routing is a future problem but today I have a far simpler one to talk about.

Word Processing.

I asked three friends to comment on a document I wrote over the weekend. I have been advocating Google Docs, (formerly known as Writely). The friends are all used email to send me the document, but no one used a word processor or highlighted their changes. The result is I am confused and want to just keep what I have. This is a bad idea in my case since my natural tendency is to right wrongly ;<). So now I am going to go into word and edit by cutting and pasting everyone's solution into one doc. Maybe this will be solved when HTML 5 brings us a collaborative environment that makes the docs blend. Until that point, my sense is that Collaboration has to be dictated from a top level domain (A.K.A. document owner or Corporate Policy).

Meritocracy and Skype

The pressure on me these days to blog is goofy.

While I am a personality (good and bad) by in large I keep my negative opinions to myself.

The Skype suit by Joltid, brings home the fact that for so many years, no one in the meritocracy has enabled enough innovation to overcome these set backs.

All too often the battles goals at the IETF are thwarted by a general acceptance of the status quo and the need to move to bigger fires.

My hope is that the Joltid suit will awaken some of the P2P efforts at the IETF to enable an alternative.

The Father of SIP on the Perils of P2P.

Henning Schulzrinne was at Illinois Institute of Technology VoIP event last week.  (It would have been great to have him in Boston, but alas.)

A few years ago at a SIP Summit I ran I thought I understood him to say that Peer 2 Peer had scaling problems, this conference I understood why.

First of all, as any good discussion he defined the term peer 2 peer and explained that a peer 2 peer system had to have these attributes.

  • Each peer must act as both a client and a server
  • Peers provide computational or storage resources for other peers
  • Self-organizing and scalling

They also Include the ability to

  • Traverse NATS,
  • replicate the service and
  • provide data storage.

He pointed out that P2P systems are not causing backbone congestion as has been widely publicized the backbone has stayed a steady 20% utilization for P2P systems. P2P systems are great for building new services that have little infrastructure cost and great flexibility.

He then looks at the issues of the cost of power and processing and determined that Client server is greener.  For security, the system is more vulnerable and for network management harder to troubleshoot.
Other issues being faced are the unfriendly nature of mobile environments and video streaming.

Henning’s most interesting comment was that P2P was probably going to take over some home and office based solutions.  I also got the sense that a P2P network management client could be a very interesting project.
It was a very interesting discussion. And I recommend tracking Henning and his student’s activities.

P4P Interview with Doug Pasko of Verizon

With the IETF meetings over in Dublin, I am expecting to have a few items of discussion coming up in the near future about standards.

New strategies for Peering Networks are being discussed in Distributed Computing Industry Association and Verizon and some other carriers are looking to implement the P4P strategies best articulated in the article posted here. P4P, which has a lengthy name explanation, should be called either Peering 4 Providers or Providers 4 Peering, imho.

Doug Pasko, PMTS – Principal Member of Technical Staff at Verizon Network & Technology (who is also the co-chair of the DCIA’s P4P group with Laird Popkin of Pando Networks) ,
does a great job of explaining to me to the value of P4P. In this interview, Doug gives an overview of the solution and why it has value to a network operator and the application provider.