Remember, every device (router, computer, smartphone, smartwatch, smart TV, smart speaker, etc.) connected to the Internet will have its own IP address and there are nearly 8 billions human beings on planet Earth. Though this seems like a large number, we are reaching a stage where we are running out of unique IP addresses. Some devices can be configured on an IP network to have a static IP address: it will remain she same even if you restart your device.Īs an IPv4 address consists of 4 Bytes (=32 bits), there are 2 32 possible IP addresses which is just under 4.3 billions IP addresses. Note the g bit stays untouched as 0 for link-local addresses.A dynamic IP Address can change every time your restart your device/reconnect to the network. For example if you have 00.11.22.33.44.55 you would change the 00 to 02 since the u bit is set to 1 and between 22 and 33 you would insert FF.FE which results in 02.11.22.FF.FE.33.44.55. The process involved setting the u bit to 1 to represent locally administered and inserting FFFE in the middle of the 6 bytes of EUI-48. Now the modified EUI-64 format is a EUI-48 that’s been modified into a EUI-64 type IID. The u bit represents if it’s universally administered (0) or locally administered (1) and the g bit represents if it’s an individual (0) or group/multicast-type (1) address. In both EUI-64 and EUI-48 format you will have the u bit and g bit as show in the figure below. Many IEEE standard compliant interfaces like Ethernet use EUI-48 format (MAC addresses) where you only have 24 bits after the OUI. IIDs are typically formed from the link-layer MAC address via modified EUI-64 format or a randomized process typically seen in Apple operating systems today.ĮUI-64 (non-modified method) consists of a 24 bit OUI allocated by IEEE-RA followed by 40 bit extended identifier typically set by the manufacturer/organization. Link-local and some global addresses use IIDs in the last 64 bit portion of the IPv6 address. OUI = Organizationally Unique Identifier (maintained by IEEE registration authority) In both the IPv4 and IPv6 multicast cases, the "G" bit is 1 in the corresponding 48-bit multicast ethernet MAC. The IPv4 groups start off with the IETF's special IEEE 0x01005e prefix and the IPv6 groups with the similarly allocated 0x3333. Packet captures will show this on Linux so will " ip maddr show". The multicast bit carries over to layer 2 for Ethernet compare the on-link MAC addresses of your multicast group memberships from your favorite client with its unicast addresses to highlight the difference. Since most organizations aren't yet sending globally routed IPv6 multicast the G bit is usually 0. The bit labelled "G" is the individual/group bit it is 0 for unicast addresses and 1 for multicast addresses. I haven't read the Microsoft book was the context of your quote 48-bit layer-2 ethernet addresses or 64-bit layer 3 IPv6 mapped host parts? You should get different values for U/L depending on where you are using it. The IETF significance of the EUI-64 U/L bit is backwards from the EUI-48 significance. Otherwise 0, which allows static locally assigned addreses such as 2001:db8::53 to be written conveniently. The general idea seems to be that if the host part is based on either a raw IEEE EUI-64 identifier or a mapped EUI-48 identifier (think ethernet MAC), it should be globally unique, so the IETF U/L bit is set to 1. In a better world, perhaps :-( The footnote to appendix A in RFC 4291 indicates that even the standards writers can't get it right. Surely this is straightforward from the specification?
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |