News for hackers. Stuff that matters.
Re:New PRISM slides say the program allows NSA to eavesdrop ... by mpyne on Sunday June 30, 2013 @11:25AM (#5966059)
> means that there is nothing between the servers and the wiretapping equipment.

There can easily still be no wiretapping equipment upstream of the servers; the companies are already copying the data directly to the collection equipment anyways, so that can be a parallel path. I'm not sure a place like Google or Facebook would ever choose to participate in something like this if their whole business were one misconfigured serial-path wiretap server away from being completely down.

I believe the NYT had mentioned a couple of days after the story broke that 'real-time' collection was possible. Even that is easy technically, you just send a copy of that instant message or video frame to the FBI collection server as it goes out.

Re:New PRISM slides say the program allows NSA to eavesdrop ... by yk on Sunday June 30, 2013 @12:00PM (#5966190)
By 'nothing between' I meant 'nothing that is worth mentioning in a newspaper article.' But you raise an interesting point, the entire wiretapping needs to handle a similar load as a large datacenter and should not impact the uptime, and therefore it should be quite an engineering challenge. So there should be documentation on the Google side, how to 'plug' it into a data center, which power requirements that stuff has, etc...
Re:New PRISM slides say the program allows NSA to eavesdrop ... by mpyne on Sunday June 30, 2013 @02:47PM (#5966711)
Which means that it's not something that Larry Page can just sneak into his datacenter without his squads and squads of Googlers never picking up on it, true.
Re:New PRISM slides say the program allows NSA to eavesdrop ... by Terretta on Sunday June 30, 2013 @11:37AM (#5966095)
Even the cheapest managed switches have a traffic mirroring port setting, no "prism" required.
Re:New PRISM slides say the program allows NSA to eavesdrop ... by mpyne on Sunday June 30, 2013 @11:45AM (#5966131)
Even if we assume that PRISM still needs to know what exactly to do with that inbound traffic, which it will need metadata for.

Of course maybe it's possible to setup a session and then just forward all traffic from a given company subnet IP:port to a PRSIM case file, but that's getting back into impl details for me.