Study of Packet Level UDP Performance of NAT44, NAT64 and IPv6 Using Iperf in the Context of IPv6 Migration

Abstract

The Internet Assigned Number Authority (IANA) allocated the last of the available /8's of the IPv4 address space to the Regional Internet Registries (RIR's) on February 2011. It could not be denied that IPv6 is the Internet of the next generation, however its utilization and implementation in a wide scale had brought hesitation to the users since it will take time and there are concerns that need to be explored in the future. Hence, this paper will lead the way for the acceptance of Internet Protocol version 6 (IPv6) migration in the Philippines using a similar Network Address Translation (NAT) that there is an apparent means to be taken into consideration and NAT IPv6 to IPv4 (NAT64) can be a good choice for computer networks with the similar situation like the Philippines which is behind NAT44, showing service providers and institutions in putting NAT64 as a viable option to be used. Therefore it is concluded, where overall performance revealed that IPv6 and NAT64 network offered better performance against the NAT44 network in almost all of the metrics in the iperf generic UDP mode test. For time transfer, IPv6 had 19% less and NAT64 with 31% less where both executed the transfer in a lesser time compared to NAT44 networks with 48% which is longer than the results of the aforementioned networks. In the bandwidth utilization, it is presented that IPv6 network had 43% and NAT64 network with 41% offered better bandwidth utilization as compared to NAT44 network with 15%. IPv6 network had 49% showed faster transfer rate along with NAT64 network with 39% in comparison to the NAT44 network with only 10% transfer. For jitter, IPv6 network had lesser average with 28% and so with NAT64 network with 31% compared to NAT44 with 40% jitter which is a quite bigger than the other experimental networks. For packet loss, IPv6 network had 26% and NAT64 network with 33% compared to NAT44 network that had 40% bigger than the other experimental networks.