DS-Lite from what I've read is no better than CGN in the sense that something still has to translate the IPv6 ip of the customer to a IPv4 address from the pool of available addresses all the while keeping a tunnel open to the IPv6 endpoint (CPE). This may be a better solution than whatever else is available, due to the lack of movement on switching to IPv6 any ISP has the choice between llama-goat-crap and wow-holy-bovine-crap. DS-lite pretty much also assumes that the customer only wants to make connections out via IPv4, with no inbound connections allowed. There is almost no way to have a 1-to-1 mapping between IPv4 to IPv6 (any ISP with enough address space available to have a DS-lite IPv4 pool that big will just run dual-stack).
Also based on http://www.networkworld.com/co...
If a simple mapping between inside IPv4 source address / port to outside IPv4 source address / port was performed on outgoing packets, as is done with regular NAT44, the LSN would have no way to differentiate between overlapping RFC1918 IPv4 addresses in different customer networks.
In other words the LSN has to somehow be able to differentiate between 192.168.1.5 on your network (which might be your PS4 but for the guy down the street its his wife's laptop). This is normally handled by VRF (separate routing / arp / NAT table) per customer, Thankfully they have dealt with this by just tacking on the customer's unique IPv6 address to the record it just makes what I expect to be huge NAT tables even larger. The diagrams from that article also show that the real benefits of DS-lite won't start showing up until the end user's devices are running IPv6 natively (only then can they take advantage of the direct paths, instead of the translated paths).
So if I'm understanding you (and DS-Lite) correctly, how does this remove the need for at least some part of the service provider to understand both IPv6 and IPv4? To me it concentrates the load on the translator devices in exchange for removing the need for the entire network to understand IPv4. In the short term this will be an extremely high load for these devices to maintain, I guess the hope is only token effort has to be put into them so it forces users to switch to IPv6 when available. Given that only 3 of the top 10 sites on http://www.alexa.com/topsites lack IPv6 records (twitter, amazon and baidu) that may not be an unreasonable expectation (the heavy streaming sites like, youtube & netflix are IPv6 so load may actually be lower than expected).