In the literature study, it is observed that the first
category, dual stack mechanism needs both end systems and
network devices to be IPv4 and IPv6 enabled. Further, they
need two IP stacks, must maintain two processing tables and
require more memory. Moreover, IPv4 address must be
available for every dual-stack machine. This is annoying,
since IPv6 was developed precisely due to the scarcity of
IPv4 addresses. Applications must be able to determine
whether the peer is using IPv4 or IPv6. The second category,
tunneling mechanism increases packet overhead and the
border routers need to be dual stacked.
The third category, translation mechanism, both SIIT and
NAT provide one way translation where IPv4 nodes cannot
initiate communications with IPv6 nodes. NAT-PT ALG is
required to carry IP address in the higher layers; NAPT-PT
allows a set of IPv6 nodes to share only a single IPv4 address.
TRT can be used as a temporary transition to allow IPv4 and
IPv6 to interoperate. SOCKS provides only unidirectional
communication. BDMS needs to have additional
components like DNS46 server and v4-v6 enabled gateway.
Virtual Overlay architecture results in a large header
overhead due to the three encapsulations needed to transport
the IPv6 packets HAv6; does not consider the continuity of
the IPv4 service. In MIP-ALG, if the IP versions of the HA
and FA are different, more signaling between NAT-PT and
HA and MIP-version translations are required, which results
in registration delay. DSMIPv6 suffers in the case of IPSec
encryption, which is mandatory for signaling messages. In
Residential Gateway architecture, if the arrival of number of
packets exceeds the queue size, there may be packet loss.
Table (I) below summarizes all three categories of various
transition implemented architectures.