IP adalah protokol yang menyalurkan datagram IP ke saluran yang tertentu
dengan harapan ia sampai ke destinasinya. IP adalah protokol yang
fleksibel. Ia boleh menyalurkan datagram IP ke laluan yang berbeza
bergantung kepada keadaan pada laluan yang ada. Penghala yang bijak dan
canggih boleh menentukan laluan yang terdekat untuk sampai ke destinasi
dan sebagainya. IP dikatakan tidak reliabl kerana ia tidak mengambil
tahu sama ada bungkusan komunikasi sampai ke destinasinya atau tidak.
Bagaimanapun, IP adalah tunggak kepada TCP/IP kerana setiap protokol di
lapisan atas dan di bawah menggunakan IP.
Dengan itu saya ingin berkongsi mengenai perbezaan IPv4 dengan IPv6:
| Description | IPv4 | IPv6 |
|---|---|---|
| Address | 32 bits long (4 bytes). Address is composed
of a network and a host portion, which depend on address class. Various
address
classes are defined: A, B, C, D, or E depending on initial few bits.
The
total number of IPv4 addresses is 4 294 967 296. The text form of
the
IPv4 address is nnn.nnn.nnn.nnn, where
0<=nnn<=255,
and
each n is a
decimal digit. Leading
zeros can be omitted. Maximum number of print characters is 15, not
counting
a mask. |
128 bits long (16 bytes). Basic architecture
is 64 bits for the network number and 64 bits for the host number.
Often,
the host portion of an IPv6 address (or part of it) will be derived from
a
MAC address or other interface identifier. Depending on the subnet
prefix,
IPv6 has a more complicated architecture than IPv4. The number of IPv6 addresses is 1028 (79 228 162 514 264 337 593 543 950 336) times larger than the number of IPv4 addresses. The text form of the IPv6 address is xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx, where each x is a hexadecimal digit, representing 4 bits. Leading zeros can be omitted. The double colon (::) can be used once in the text form of an address, to designate any number of 0 bits. For example, ::ffff:10.120.78.40 is an IPv4-mapped IPv6 address. (See  RFC 3513 for details. To view this RFC, see RFC Editor 
(www.rfc-editor.org/rfcsearch.html). |
| Address allocation | Originally, addresses were allocated by network class. As address space is depleted, smaller allocations using Classless Inter-Domain Routing (CIDR) are made. Allocation has not been balanced among institutions and nations. | Allocation is in the earliest stages. The Internet Engineering Task Force (IETF) and Internet Architecture Board (IAB) have recommended that essentially every organization, home, or entity be allocated a /48 subnet prefix length. This would leave 16 bits for the organization to do subnetting. The address space is large enough to give every person in the world their own /48 subnet prefix length. |
| Address lifetime | Generally, not an applicable concept, except for addresses assigned using DHCP. | IPv6 addresses have two lifetimes: preferred
and valid, with the preferred lifetime always <= valid. After
the preferred lifetime expires, the address is not to be used as a
source
IP address for new connections if an equally good preferred address is
available.
After the valid lifetime expires, the address is not used (recognized)
as
a valid destination IP address for incoming packets or used as a source
IP
address. Some IPv6 addresses have, by definition, infinite preferred and valid lifetimes; for example link-local (see address scope). |
| Address mask | Used to designate network from host portion. | Not used (see address prefix). |
| Address prefix | Sometimes used to designate network from host portion. Sometimes written as /nn suffix on presentation form of address. | Used to designate the subnet prefix of an address. Written as /nnn (up to 3 decimal digits, 0 <= nnn <= 128) suffix after the print form. An example is fe80::982:2a5c/10, where the first 10 bits comprise the subnet prefix. |
| Address Resolution Protocol (ARP) | Address Resolution Protocol is used by IPv4 to find a physical address, such as the MAC or link address, associated with an IPv4 address. | IPv6 embeds these functions within IP itself as part of the algorithms for stateless autoconfiguration and neighbor discovery using Internet Control Message Protocol version 6 (ICMPv6). Hence, there is no such thing as ARP6. |
| Address scope | For unicast addresses, the concept does not apply. There are designated private address ranges and loopback. Outside of that, addresses are assumed to be global. | In IPv6, address scope is part
of the architecture. Unicast addresses have two defined scopes,
including
link-local and global; and multicast addresses have 14 scopes. Default
address
selection for both source and destination takes scope into account. A scope zone is an instance of a scope in a particular network. As a consequence, IPv6 addresses sometimes must be entered or associated with a zone ID. The syntax is %zid where zid is a number (usually small) or a name. The zone ID is written after the address and before the prefix. For example, 2ba::1:2:14e:9a9b:c%3/48. |
| Address types | Unicast, multicast, and broadcast. | Unicast, multicast, and anycast. See IPv6 address types for descriptions. |
| Communications trace | A tool to collect a detailed
trace of TCP/IP (and other) packets that enter and leave the system. |
Same for IPv6, and IPv6 is
supported. |
| Configuration | You must configure a newly installed system
before it can communicate with other systems; that is, IP addresses and
routes
must be assigned. |
Configuration is optional,
depending on functions required. IPv6 can be used with any Ethernet
adapter
and can be run over the loopback interface. IPv6 interfaces are
self-configuring
using IPv6 stateless autoconfiguration. You can also manually configure
the
IPv6 interface. So, the system will be able to communicate with other
IPv6
systems that are local and remote, depending on the type of network and
whether
an IPv6 router exists. |
| Domain Name System (DNS) | Applications accept host names and then use
DNS to get an IP address, using socket API gethostbyname().
Applications also accept IP addresses and then use DNS to get host
names
using gethostbyaddr(). For IPv4, the domain for reverse lookups is in-addr.arpa. |
Same for IPv6. Support for IPv6 exists using
AAAA (quad A) record type and reverse lookup (IP-to-name). An
application
may elect to accept IPv6 addresses from DNS (or not) and then use IPv6
to
communicate (or not).
The socket API gethostbyname()
only
supports IPv4. For IPv6, a new getaddrinfo()
API is used
to obtain (at application choice) IPv6 only, or IPv4 and IPv6 addresses. For IPv6, the domain used for reverse lookups is ip6.arpa, and if not found then ip6.int (see API getnameinfo()). |
| Dynamic Host Configuration Protocol (DHCP) | Used to dynamically obtain
an IP address and other configuration information. i5/OS
supports
a DHCP server for IPv4. |
The i5/OS implementation of DHCP does not
support IPv6. |
| File Transfer Protocol (FTP) | File Transfer Protocol allows you to send and receive files across networks. | The i5/OS implementation of FTP does not support IPv6. |
| Fragments | When a packet is too big for the next link over which it is to travel, it can be fragmented by the sender (host or router). | For IPv6, fragmentation can
only occur at the source node, and reassembly is only done at the
destination
node. The fragmentation extension header is used. |
| Host table | On iSeries Navigator, a configurable table that associates an Internet address with a host name; for example, 127.0.0.1, loopback. This table is used by the sockets name resolver, either before a DNS lookup or after a DNS lookup fails (determined by host name search priority). | Currently, this table does not support IPv6. Customers need to configure an AAAA record in a DNS for IPv6 domain resolution. You can run the DNS locally on the same system as the resolver, or you can run it on a different system. |
| Interface | The conceptual or logical entity used by
TCP/IP to send and receive packets and always closely associated with an
IPv4
address, if not named with an IPv4 address. Sometimes referred to as a
logical
interface. Can be started and stopped independently of each other
and
independently of TCP/IP using STRTCPIFC and ENDTCPIFC commands and using
iSeries Navigator. |
Same concept as IPv4. Can be started and
stopped independently of each other and independently of TCP/IP using
iSeries Navigator
only. |
| Internet Control Message Protocol (ICMP) | ICMP is used by IPv4 to communicate network information. | Used similarly for IPv6; however, Internet
Control Message Protocol version 6 (ICMPv6) provides some new
attributes.
Basic error types remain, such as destination unreachable, echo
request
and reply. New types and codes are added to support neighbor discovery
and
related functions. |
| Internet Group Management Protocol (IGMP) | IGMP is used by IPv4 routers to find hosts that want traffic for a particular multicast group, and used by IPv4 hosts to inform IPv4 routers of existing multicast group listeners (on the host). | Replaced by MLD (multicast listener discovery) protocol for IPv6. Does essentially what IGMP does for IPv4, but uses ICMPv6 by adding a few MLD-specific ICMPv6 type values. |
| IP header | Variable length of 20-60 bytes, depending on IP options present. | Fixed length of 40 bytes. There are no IP header options. Generally, the IPv6 header is simpler than the IPv4 header. |
| IP header options | Various options that might accompany an IP header (before any transport header). | The IPv6 header has no options. Instead, IPv6 adds additional (optional) extension headers. The extension headers are AH and ESP (unchanged from IPv4), hop-by-hop, routing, fragment, and destination. Currently, IPv6 supports some extension headers. |
| IP header protocol byte | The protocol code of the transport layer or packet payload; for example, ICMP. | The type of header immediately following the IPv6 header. Uses the same values as the IPv4 protocol field. But the architectural effect is to allow a currently defined range of next headers, and is easily extended. The next header will be a transport header, an extension header, or ICMPv6. |
| IP header Type of Service (TOS) byte | Used by QoS and differentiated services to designate a traffic class. | Designates the IPv6 traffic class, similarly to IPv4. Uses different codes. Currently, IPv6 does not support TOS. |
| iSeries Navigator support | iSeries Navigator provides a complete configuration solution for TCP/IP. | Same for IPv6. No CL commands are available for IPv6 configuration. |
| LAN connection | Used by an IP interface to get to the
physical network. Many types exist; for example, token ring, and
Ethernet.
Sometimes referred to as the physical interface, link, or line. |
IPv6 can be used with any Ethernet adapters and
is also
supported over virtual Ethernet between logical partitions. |
| Layer 2 Tunnel Protocol (L2TP) | L2TP can be thought of as virtual PPP, and works over any supported line type. | Currently, the i5/OS implementation of L2TP does not support IPv6. |
| Loopback address | An interface with an address of 127.*.*.* (typically 127.0.0.1) that can only be used by a node to send packets to itself. The physical interface (line description) is named *LOOPBACK. | The concept is the same as in IPv4. The single
loopback address is 0000:0000:0000:0000:0000:0000:0000:0001
or ::1 (shortened
version). The virtual physical interface is named *LOOPBACK. |
| Maximum Transmission Unit (MTU) | Maximum transmission unit of a link is the maximum number of bytes that a particular link type, such as Ethernet or modem, supports. For IPv4, 576 is the typical minimum. | IPv6 has an architected lower bound on MTU of 1280 bytes. That is, IPv6 will not fragment packets below this limit. To send IPv6 over a link with less than 1280 MTU, the link-layer must transparently fragment and defragment the IPv6 packets. |
| Netstat | A tool to look at the status of TCP/IP connections, interfaces, or routes. Available using iSeries Navigator and 5250. | Same for IPv6, and IPv6 is supported for both 5250 and iSeries Navigator. |
| Network Address Translation (NAT) | Basic firewall functions integrated into TCP/IP, configured using iSeries Navigator. | Currently, NAT does not support IPv6. More generally, IPv6 does not require NAT. The expanded address space of IPv6 eliminates the address shortage problem and enables easier renumbering. |
| Network table | On iSeries Navigator, a configurable table that associates a network name with an IP address without mask. For example, host Network14 and IP address 1.2.3.4. | Currently, no changes are made to this table for IPv6. |
| Node info query | Does not exist. | A simple and convenient network tool that should work like ping, except with content: an IPv6 node may query another IPv6 node for the target's DNS name, IPv6 unicast address, or IPv4 address. Currently, not supported. |
| Packet filtering | Basic firewall functions integrated into TCP/IP, configured using iSeries Navigator. | You
cannot use packet filtering with IPv6. |
| Packet forwarding | The i5/OS TCP/IP
stack can be configured to forward IP packets it receives for nonlocal
IP
addresses. Typically, the inbound interface and outbound interface are
connected
to different LANs. |
IPv6 packets are not forwarded. |
| PING | Basic TCP/IP tool to test reachability. Available using iSeries Navigator and 5250. | Same for IPv6, and IPv6 is supported, for both 5250 and iSeries Navigator. |
| Point-to-Point Protocol (PPP) | PPP supports dialup interfaces over various modem and line types. | Currently, the i5/OS implementation of PPP does
not support
IPv6. |
| Port restrictions | These i5/OS panels
allow a customer to configure a selected port number or port number
ranges
for TCP or UDP so that they are only available for a specific profile. |
Same for IPv6. Port restrictions
for IPv6 are identical to those available in IPv4. |
| Ports | TCP and UDP have separate port spaces, each identified by port numbers in the range 1-65535. | For IPv6, ports work the same as IPv4. Because these are in a new address family, there are now four separate port spaces. For example, there are two TCP port 80 spaces to which an application can bind, one in AF_INET and one in AF_INET6. |
| Private and public addresses | All IPv4 addresses are public, except for three address ranges that have been designated as private by IETF RFC 1918: 10.*.*.* (10/8), 172.16.0.0 through 172.31.255.255 (172.16/12) , and 192.168.*.* (192.168/16). Private address domains are commonly used within organizations. Private addresses cannot be routed across the Internet. | IPv6 has an analogous concept, but with important
differences. Addresses are public or temporary, previously termed
anonymous.
See RFC 3041. Unlike IPv4 private addresses, temporary addresses can be
globally
routed. The motivation is also different; IPv6 temporary addresses are
meant
to shield the identity of a client when it initiates communication (a
privacy
concern). Temporary addresses have a limited lifetime, and do not
contain
an interface identifier that is a link (MAC) address. They are generally
indistinguishable
from public addresses. IPv6 has the notion of limited address scope using its architected scope designations (see address scope). |
| Protocol table | On iSeries Navigator, a configurable table that associates a protocol name with its assigned protocol number; for example, UDP, 17. The system is shipped with a small number of entries: IP, TCP, UDP, ICMP. | The table can be used with
IPv6 without change. |
| Quality of service (QoS) | Quality of service allows you to request packet priority and bandwidth for TCP/IP applications. | Currently, the i5/OS implementation
of QoS does not support IPv6. |
| Renumbering | Done by manual reconfiguration, with the possible exception of DHCP. Generally, for a site or organization, a difficult and troublesome process to avoid if possible. | Is an important architectural element of IPv6, and is largely automatic, especially within the /48 prefix. |
| Route | Logically, a mapping of a set of IP addresses
(might contain only one) to a physical interface and a single next-hop
IP
address. IP packets whose destination address is defined as part of the
set
are forwarded to the next hop using the line. IPv4 routes are associated
with
an IPv4 interface, hence, an IPv4 address. The default route is
*DFTROUTE. |
Conceptually, similar to IPv4.
One important difference: IPv6 routes are associated (bound) to a
physical
interface (a link, such as ETH03) rather than an interface. One reason
that
a route is associated with a physical interface is because source
address
selection functions differently for IPv6 than for IPv4. See Source
address
selection. |
| Routing Information Protocol (RIP) | RIP is a routing protocol supported by the routed daemon. | Currently, RIP does not support IPv6. IPv6 routing uses static routes. |
| Services table | On i5/OS,
a configurable table that associates a service name with a port and
protocol;
for example, service name FTP-control, port 21, TCP, and User Datagram
Protocol
(UDP). A large number of well-known services are listed in the services table. Many applications use this table to determine which port to use. |
No changes are made to this table for IPv6. |
| Simple Network Management Protocol (SNMP) | SNMP is a protocol for system management. | Currently, the i5/OS implementation
of SNMP does not support IPv6. |
| Sockets API | These APIs are the way applications use TCP/IP. Applications that do not need IPv6 are not affected by sockets changes to support IPv6. | IPv6 enhances sockets so that applications
can now use IPv6, using a new address family: AF_INET6. The
enhancements
have been designed so that existing IPv4 applications are completely
unaffected
by IPv6 and API changes. Applications that want to support concurrent
IPv4
and IPv6 traffic, or IPv6-only traffic, are easily accommodated using
IPv4-mapped
IPv6 addresses of the form ::ffff:a.b.c.d,
where a.b.c.d is
the IPv4 address of the client. The new APIs also include support for converting IPv6 addresses from text to binary and from binary to text. See Using AF_INET6 address family for more information about sockets enhancements for IPv6. |
| Source address selection | An application may designate a source IP (typically, using sockets bind()). If it binds to INADDR_ANY, a source IP is chosen based on the route. | As with IPv4, an application can designate a source IPv6 address using bind(). Similarly to IPv4, it can let the system choose an IPv6 source address by using in6addr_any. But because IPv6 lines have many IPv6 addresses, the internal method of choosing a source IP is different. |
| Starting and stopping | Use STRTCP and ENDTCP to start or end TCP/IP. | Same as IPv4. IPv4 and IPv6 are not started
or stopped independently of one another or independently of TCP/IP. That
is,
you start and stop all of TCP/IP, not just IPv4 or IPv6. Any
IPv6 interfaces are automatically started if the AUTOSTART parameter = *YES (the
default). IPv6 cannot be used or configured without IPv4. The IPv6
loopback
interface, ::1, will automatically be
defined and activated
when IPv6 is started. |
| Telnet | Telnet allows you to log on and use a remote computer as though you were connected to it directly. | Currently, the i5/OS implementation
of Telnet does not support IPv6. |
| Trace route | Basic TCP/IP tool to do path determination. Available using iSeries Navigator and 5250. | Same for IPv6, and IPv6 is supported for both 5250 and iSeries Navigator. |
| Transport layers | TCP, UDP, RAW. |
The same transports exist in IPv6. |
| Unspecified address | Apparently, not defined, as such. Socket programming uses 0.0.0.0 as INADDR_ANY. | Defined as ::/128 (128 0 bits). It is used as the source IP in some neighbor discovery packets, and various other contexts, like sockets. Socket programming uses ::/128 as in6addr_any. |
| Virtual private network (VPN) | Virtual private network (using IPsec) allows you to extend a secure, private network over an existing public network. | Currently, the i5/OS implementation of VPN does
not support
IPv6. |
