On Thu, 2010-07-08 at 13:54 +0200, Roland Bless wrote: > Hi Steven, > > On 07.07.2010 18:57, Steven Blake wrote > >> Do you think ILNP could be reliable and robust if the Identifiers > >> used by hosts are not in actual fact globally unique? If so, please > >> give some examples. > > > > I think they need to be unique on a subnet. Being globally unique with > > high probability makes uniqueness on a subnet easier to achieve. > > I don't think that ILNP will work out of the box if two > nodes have the same ID but are located in different subnets. > The transport cannot distinguish whether it's a connection > to itself or to the other remote system if it only operates on the ID. > I think that the problem of conflicting IDs but different Locators > is probably solvable, but I haven't seen any proposal for this so far.
No one ever said explictly that you cannot use locators as part of the socket demux (directly, or indirectly, by mapping remote locator +identifier to a correspondent cache entry whose internal index is used in socket demux). Locators aren't used in the transport checksum. I previously sent Ran a comment to clarify this, FWIW. > > Two hosts on the same link with the same MAC address can't communicate > > using any protocol (IPv4, IPv6, ILNP, etc.). Fortunately, the > > occurrence of duplicate MACs is extremely rare. > > > > Name one modern device that doesn't have a unique hardware ID that could > > be used to form a global-scope EUI-64? > > Hmm, when considering the increasing use of virtual hosts, we > may get in trouble here: virtual hosts probably don't have a unique > hardware ID and usually generate a MAC address (or multiple ones > in case of several virtual interfaces) at installation time. VMs usually have their MAC addresses assigned centrally by a Virtual Machine Manager. VMware vCenter does this out of a VMware's own OUI pool. If you are setting up VMs by hand, you have got to be careful, or you will break bridging! > > Nothing breaks if two hosts on separate subnets share the same ID, even > > if a separate host is communicating simultaneously with both of them. > > As long as you have the Locator in addition to the ID available, > that's ok, but as far as I understood, the transport layer only > uses I. So further demultiplexing will only work if the transport > layer considers L:I together. Is this the case somewhere, e.g., when > passing packets to the network layer? Basically correct. The definition of the boundary between ILNP network layer and transport layer is left deliberately vague, because the are multiple viable implementation alternatives. For example, Christian Vogt's name-based sockets could be used within transport, and the name mapped to some internal correspondent cache entry at the transport/network-layer boundary. Regards, // Steve _______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
