I have been asked to review this document on behalf of the ops directorate. This document is intended for experimental status, and describes a number of strategies to take when performing EDNS(0) padding. It recommends one (experimental) option to use based on empirical data. Overall, I think this document is close to be ready. In general, coming at it from an operator standpoint, I thought the layout was a bit odd. The recommended option is spelled out first, but then the document goes into sub-optimal approaches before it actually describes the full recommended solution (both from a query and response stance). It might flow better to discuss the recommended approach in detail while leaving all of the sub-optimal approaches for appendices. At the very least, the Maximal Length Padding approach feels somewhat non-sensible and should be pushed to the appendix. Nit-wise, I found the following: Section 3: s/signifcantly/significantly/ s/octects/octets/ === Section 4.1 You refer to Maximum Length Padding here, but Section 4.2 calls it "Maximal Length Padding". === Section 4.2 Is referencing "The Full Monty" needed here? === Section 4.3 You expand (pseudo) to (pseudo) random number in Section 4.1. I think the same should be done here for clarity. === Section 4.4 s/transction/transaction/ === Section 5 When you talk about a multiple of 468 bytes used for response padding, I think you should include a "see below". === Section 5 s/octects/octets/ === Section 5 Where you have "Note that the recommendation above does apply only..." I think it reads better to say, "Note that the recommendation above applies only..." === Section 8 s/inffective/ineffective/