Nothing new in this mail but I need some clarification ... - To my understanding, the common practice is, at least, to accept any address block larger or equal to /19 and to filter any block smaller than /24 (except eventually at the customer-ISP edge). ARIN and RIPE have changed for a while their minimum allocation size to /20 so I guess there is no risk of routing blackhole with /20 blocks, right? - What about PI-blocks between 21 and 24 bits? If a end-user network annouces such an address block, what is the risk of routing blackholes throughout the Internet? - Each ISP seems to apply its own filtering policy. Nevertheless, what are the most common filtering practices for the ISP-ISP edge? - Is it common to have different filtering policies depending on the address ranges? I will be grateful for any link to official or non-official documentation. Thanks for your help Alex