The article is wrong about address assignments. First, /32 is the minimum allocation size, which means you (an ISP) can't get fewer addresses than that (not mo

The article is wrong about address assignments. First, /32 is the minimum allocation size, which means you (an ISP) can't get fewer addresses than that (not more as the article claims). The reason you'd want to have a minimum allocation size is because it makes routing tables smaller (better for an organization to have just one entry in the routing table that's huge that they can grow into, than to start small and add lots of small entries to the routing table as they grow). Second, /48 is what ISPs typically give their subscribers, not what ISPs themselves get. Edit: to clarify, the /32 minimum is a policy of ARIN and RIPE (not sure about the other RIRs). See for example https://www.arin.net/policy/archive/ipv6_policy.html#43 https://www.arin.net/policy/archive/ipv6_policy.html#43 and http://www.ripe.net/ripe/docs/ripe-552#minimum_allocation http://www.ripe.net/ripe/docs/ripe-552#minimum_allocation. There is no mandatory policy for how ISPs should assign addresses to their subscribers. RFC3177 recommended a /48 in most cases, though that has recently been superseded by RFC6177, which recommends more flexibility rather than a one-size-fits all approach. See http://tools.ietf.org/html/rfc6177 http://tools.ietf.org/html/rfc6177. I find the article somewhat reckless in recommending that you ban entire /64s instead of individual IPv6 addresses. It's true that you need to be aware that a home user will likely be in control of an entire /64 (and possibly more), but if the offending IP address is at a university or a datacenter then a /64 ban could sweep up a lot of innocent bystanders. You really need to consider bans on a case-by-case basis.