When transitioning metropolitan area networks to IPv6, two critical technologies stand out: NAT444 and DS-Lite. What fundamental factors should you consider when choosing between them? This blog post details the principles behind both solutions and how to approach post-deployment.
Metropolitan area networks have a huge number of broadband users, making them the primary consumers of IPv4 addresses. Facing IPv4 address exhaustion, one countermeasure is to slow down IPv4 consumption by moving away from each user having a dedicated public IPv4 address and instead using NAT technology to enable multiple users to share a single address. The second countermeasure is to vigorously develop IPv6 applications, attracting users to the IPv6 network and reducing IPv4 consumption. The latter involves a collective linkage across the industry chain鈥擨Pv6 terminals, services, operations, and networks. Although it has developed rapidly in recent years, it remains minuscule compared to the vast IPv4 installed base. Therefore, slowing down IPv4 address consumption has become the top priority for current network builders.
During the IPv6 transition for metropolitan area networks, NAT444+DS and DS-Lite are the most discussed technical solutions.
NAT444+DS Technology: For IPv4 services, traffic is carried via two levels of NAT44. The first level of NAT44 occurs on the home network CPE side, mapping private user IPv4 addresses to carrier-grade private IPv4 addresses in a 1:1 mapping mode. The second level of NAT44 takes place on the network’s LSN (Large Scale NAT), mapping carrier-grade private IPv4 addresses to public IPv4 addresses in an N:1 mapping mode. For IPv6 services, traffic is carried via an end-to-end IPv6 protocol stack.
DS-Lite Technology: For IPv4 services, traffic is carried via IPv4-in-IPv6 tunnels combined with NAT44. The home network CPE acts as the B4 element, handling tunnel encapsulation/decapsulation for private IPv4 traffic and serving as the endpoint of the IPv4-in-IPv6 tunnel. The network CGN device acts as the AFTR element, performing decapsulation/encapsulation of tunnel data while translating private IPv4 addresses to public IPv4 addresses in an N:1 mode. For IPv6 services, similar to NAT444, traffic is carried via an end-to-end IPv6 protocol stack.
How do you choose between these two technologies based on network conditions in a metropolitan area network, and how do you select a deployment model?
First, analyze the current state of the network.
If the target network only supports IPv4 and has insufficient capability to upgrade to an IPv4/IPv6 dual stack鈥攆or example, due to outdated equipment that cannot support software upgrades for dual-stack functionality鈥攖he principle for transformation on such a network is to maintain existing services without introducing IPv6 requirements. In this case, NAT444 technology has the advantage because it allows “IPv4 address-saving” deployment for new services without requiring the network to add the IPv6 protocol. Additionally, its requirement for the home network CPE is only NAT44 functionality, which all current CPE devices support.
If the target network can support a software upgrade to IPv6 without affecting the carrying of existing IPv4 services, the principle for transformation here is to develop IPv6 services while maintaining existing ones. Since the backbone network (primarily the metropolitan area network SR, BRAS, CR) already has IPv6 Ready capability, the backbone layer meets the requirements for both NAT444+DS and DS-Lite. Due to the vast number, varied types, and differing capabilities of home network CPE devices, the transformation of the home network is the primary factor influencing the choice between NAT444+DS and DS-Lite. DS-Lite requires the CPE to support an IPv4/IPv6 dual stack and have IPv4-in-IPv6 tunnel encapsulation and decapsulation capabilities鈥攆eatures current existing CPEs lack, being present only in newly added CPEs. For single-stack IPv4 users, NAT444 only requires the CPE to support NAT44 functionality, which all existing CPEs support. For dual-stack users, it requires IPv4/IPv6 dual-stack and NAT44 functionality, which only new CPEs possess. Therefore, NAT444 has broader applicability.
If the target network is a newly built network already fully capable of IPv6, with very high support for IPv4/IPv6 functionality鈥攚here both the backbone and home network CPEs support mainstream transition technologies鈥攖hen the network layer meets the requirements for both NAT444+DS and DS-Lite. The choice then requires analysis at the next level.
Second, analyze based on service carrying and network operations.
In terms of service carrying, NAT444+DS and DS-Lite handle IPv6 traffic identically via end-to-end IPv6 protocol carrying. For IPv4 services, NAT444 performs two NAT44 translations on IPv4 packets; the packets are still carried natively in IPv4 without additional headers. DS-Lite encapsulates IPv4 packets with an IPv6 header, adding overhead which introduces issues like MTU problems and results in slightly lower carrying efficiency. Thus, from a service carrying efficiency perspective, NAT444+DS is superior to DS-Lite.
In terms of network operations, both NAT444+DS and DS-Lite impose transformation requirements on BRAS, AAA, billing, reporting, and traceability systems. For NAT444+DS, the network carries both IPv4 and IPv6 service flows, requiring management of both types during service deployment and administration. For dual-stack users, reports for corresponding IPv4 and IPv6 service flows must also be merged. For DS-Lite, users are managed based solely on their IPv6 address; user authentication, billing, and management can all be uniformly based on the user’s IPv6 address. In this regard, DS-Lite has a clear advantage in user management. Looking at the entire network, however, IPv4 and IPv6 coexist during the transition period, so a network deploying DS-Lite will still have IPv4 users, meaning OSS management based on user IPv4 addresses is still necessary. Therefore, while DS-Lite is relatively advanced in operations management and has clear advantages in networks with a high degree of IPv6 adoption, it still cannot fully achieve single-stack IPv6 management for networks in the transition phase. NAT444+DS requires complete dual-stack service management and introduces new requirements for reporting and traceability systems.
Third, analyze based on the network evolution process.
The network transformation process inevitably follows this development path: “Pure IPv4 Network -> IPv4/IPv6 Dual Stack with IPv4 Dominant -> IPv4/IPv6 Dual Stack with IPv6 Dominant -> Pure IPv6 Network.” The primary users served by both NAT444+DS and DS-Lite are private dual-stack users. For NAT444+DS, transitioning to the later stages of IPv6 development involves simplifying the network鈥攔emoving IPv4 functionality from the backbone and removing IPv4 user management from OSS systems, etc. In contrast, DS-L