> On Feb 28, 2017, at 8:29 PM, Anoop Ghanwani <an...@alumni.duke.edu> wrote: > > I support publication of the document as an informational RFC. > > Below are my comments. > > Thanks, > Anoop > > == > > - pg 5, line 1 > What is the criteria that allow sharing the AS number? Is there a > reference?
we changed this to “use the same AS”. As explained in 4.3, using the same AS brings the update loop prevention mechanism so facilitate filtering and propagation. > - pg 6 > "This means that every new connection will be established > obliviously (memory- less) with regards to the paths chosen > before, or chosen by other nodes." > I am not sure what "chosen by other nodes" adds. I think it > can be removed. It refers to the “obliviousness” extended also to the choices that other nodes of the network could have made. > - pg 7 > "local label 1600x" -> "local label (16000 + x). > Also because of the way loopbacks are assigned, does this mean that the > number nodes that this scheme can handle is 512? May be good to mention why > this is considered a good number. the example assumes loopbacks assigned from 192.0.2/24. It gives you 255 host addresses. This is of course just illustrative. > - pg 11 > "BGP Prefix Segment 16011 then directs the packet down to Node11 along the > path (Node5, Node9, Node11)." > I think it would be worth mentioning that node 9 need not appear in this > path. In general, because of the nature of clos topologies, there is no need > to have intermediate nodes between the spine and the ToR on the way down. > (If there is, it would be good to know why.) maybe I’m missing your point but the example is baed on the illustrative topology where 9 in the shortest path but you don’t need to specify 9 in the segment list. This is base of SR explained in the architecture draft. > > Editorial I fixed the remaining editorial nits. Thanks. s. > > - some inconsistencies throughout. would be good to make them consistent. > Node1 and Node2 vs Nodes 1 and 2 vs "Node1" and "Node2" > data center, data-center, DC > > - Spell out SRGB and AIGP at first use. > > - pg 1 > "use-case use-cases" -> use-cases > > - pg 5 > "via BGP session" -> "via a BGP session." (missing 'a' and period.) > "address of it's loopback" -> "address of its loopback" > "per-flow ECMP that does not" -> "per-flow ECMP does not" > "placed on one path over others" -> "placed on one path over others." > (missing period) > " implements oblivious" -> "implements an oblivious" > > - pg 6 > "Absence of path visibility" -> "The absence of path visibility" > > - pg 7 > "Figure 2 zooms on" -> "Figure 2 zooms in on" > > - pg 8 > "an nondeterministic label" -> "a non-deterministic label" > > - pg 9 > "Referring to Figure 1Referring to Figure 1" -> "Referring to Figure 1" > > - pg 11 > "if Node7 does not support" -> "even though Node7 does not support" > > - p12 > Missing a period at the end of the first and second items in Sec 4.3. > "Attribute adverting" -> "Attribute advertising" > > - pg 14 > "let us illustrate this assuming" -> "let us illustrate this concept > assuming" > "flow to Z" -> "flow to HostZ" > "assuming A is made aware" -> "assuming HostA is made aware" > > - pg 15 > "the latter one" -> "the last one" > > - pg 16 > "monitoring network elements health" -> "monitoring network elements' > health" > "inSection 7.2" -> "in Section 7.2" > "BGP Labelled Unicast" -> "BGP Labeled Unicast" (also on pg 17) > > - pg 18 > "thanks to PHP" -> "because of PHP" > "Internet- scale" -> "Internet-scale" (extra space) > "go-to-the- Internet" -> "go-to-the-Internet" > " do not recommend to use" -> "do not recommend using" > "operation viewpoint" -> "operational viewpoint" > > - pg 19 > "allows to construct" -> "allows us to construct" > "Spine5 and Spine 8" -> Node5 and Node8 > "(e.g. ToR1's SRGB is [1000, 1999], ToR2's SRGB is [2000, 2999]...)." -> > "(e.g. ToR1's SRGB is [1000, 1999], ToR2's SRGB is [2000, 2999], ...)." -> > _______________________________________________ spring mailing list spring@ietf.org https://www.ietf.org/mailman/listinfo/spring