Erik Parker wrote:
This is an instance of the "good enough" or "best effort" phenomenan. When it works 99% or 99.9% of the time, there is hardly any incentive to make it work 99.99% of the time because the observed level of service is good enough and the next event might never come. Prepareness usually has a cost associated and unless other benefits can be realized, it's easy to just ride through the rough time.
Speaking of which.. I wish P2P had been a little bit more organized when 9/11 happened.. Trying to watch the news online, download clips, or images for those few days following.. was nearly impossible. CNN/TimeWarner should recall that their entire cluster was destroyed and they had to move back to a simplified text only page that had nothing on it.. Likewise with Foxnews, but a little bit to a lesser extent.
If P2p was built upon a little bit, putting in protocols of trust (ala certificates/signed files, etc.) it could give F5 a run for its money and lower the cost drastically of certain network designs.
You don't need trust on the protocol level to disseminate information which has a verifiable source (hash). This was discussed to death just a few days ago.
Pete
