Addressing of nodes in the network
Because ANGEL APPLICATION does not use any centralized system, creating a glue for the individual nodes seems to be a hard problem. Due to the schizophrenic nature of the internet in its current state, many hosts do not have a constant or publically reachable Internet Protocol Address (IP). Although this was part of the initial design back in the 70's, this concept somehow got sidetracked by things including but not limited to the lack of IP addresses in IPv4, NAT and dial-up connections with dynamically assigned IP addresses.
For the purposes of ANGEL APPLICATION, the requirement is that a random peer called ALICE can create local clones of the data in the repository of peer BOB. To ensure data availability and to enable automatic data recovery/distribution to from/to each peer, both ALICE's and BOB's host must be reachable over the network, e.g. have a known address. Because of the aforementioned details (no central system, no constant IP), ANGEL relies on each peer providing a DNS hostname. Peers that have a dynamically assigned IP address can choose to use a loose technology commonly referred to as dynamic DNS.
Each ANGEL APPLICATION node will essentially have to have either a constant IP address, or provide the other peers with a DNS hostname that stays constant over time. This obviously introduces an additional dependency and might be considered somewhat contrasting to having a de-centralized system. Obviously, this is schizophrenic and a problem of the internet itself.
Note: since release 0.4, the ANGEL APPLICATION has built-in support for dynamic dns services of http://dyndns.majimoto.net/ and http://www.dyndns.com/