-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 27/Jan/16 09:31, joel jaeggli wrote:
There are also of course variations on this. An an aggregation router > may have quite limited FIB, e.g. enough for customer routes yet still > have a full rib in it's control-plane, at which point it needs to > default towards devices which do have a FIB in place. assuming a single > hob peering it would be rather hard to identify this case as a customer, > though if your neighbor has an Arista mac address for example that might > be a logical conclusion.
In all our Metro-E deployments, where we have enough RIB to hold multiple full feeds, but a limited FIB for forwarding, there is no discernible difference in performance both at the routing and forwarding levels to our customers that take full BGP feeds natively from these devices. Mark. -----BEGIN PGP SIGNATURE----- iQIcBAEBCAAGBQJWqHPKAAoJEGcZuYTeKm+GgPgP/2v9PaX7cfg7re6jAhbbWkzp sw5jsPdF0eCmJAOICvv74ZymRzW8fAmlt98XWpmAJh/8WqDKkn/H1lEt1hvsyuYE 9PC3PRsT2+Qhb26Erlz1LB+95dS6PzZyNzHC6YRRbB2j3aZkazkOHCZTl4lWZeIP ZsiafWQl3LDCUOrgO4JsqVg3r4D4DhMATKxQuP5siXOiEpwVk1zWSa+MfydrUrg7 jlGzwkH1Igh1UmsMy2oSW9azjQizSyBl6/fdbx3sbCHqHrtXbAt7TSrF0kJ//1lm JHYhGM1vovpxKCOxY74AiemrSXFbkDCSc8LgiMRPL3l5VfquYVy6jXFpiPM/H9sU xKUS3uKEJ4IgIIl6URMfusWTirmPC6f7mvqwOGNn/qabU4AKq+WsPshFMfb3+9Ry v0+3/o3i/hNOc+neL6oE8mHZ24pilbKltCYFD7pPTKS8lROoXfaHv4d52FMXPMUL oqdvLtRYjSb5RXpkX5hzMzJkqKJ5oVIm+Oj+KP4ekiNGRedEgiEfAPcwJG54NQgv M8Ji8cgAfwcd2lhIZXfDg1y7N4Jl5k48C1KBJV0y6nPFjtonqrufW7PEUWX17dqq ih0pcvy4dAv81hSY9jdRJucS/Kev3xLbikRA5f8vHN4h9jH7OnXh4VNrCXnetUb5 VhtsoJ6/yiz/PKJ/zEu4 =QPdU -----END PGP SIGNATURE-----