Yet another IP aggregation tool (fast)
pfxaggr [-h] [-v|-v -v] [-c] [-na|-a0|-a1|-a2|-a3|-a4]
pfxaggr reads from STDIN prefixes in CIDR notation (v4 & v6) and outputs them sorted / aggregated / in another notation to STDOUT.
make
What do the different modes do? Let's take as example the following input:
192.0.2.0/32
192.0.2.1/32
192.0.2.0/31
192.0.2.2/31
192.0.2.7/31 (unaligned)
192.0.2.2/32
192.0.2.3/32
192.0.2.0/30
192.0.2.4/31
-na just aligns and sorts the prefixes:
192.0.2.0/30
192.0.2.0/31
192.0.2.0/32
192.0.2.1/32
192.0.2.2/31
192.0.2.2/32
192.0.2.3/32
192.0.2.4/31
192.0.2.6/31
-a0 aggregates them by "hiding" prefixes, which are covered by less specifics seen in the input. No new less specifics are added:
192.0.2.0/30
192.0.2.4/31
192.0.2.6/31
-a1 aggregates them but eventually adds new less specifics. The covered address space remains the same, but added less specifics might not had been seen in the input:
192.0.2.0/29
-a2 combines the prefixes of the input in a slightly different notation for exact match with the input. prefix/mask-maxmask states, that in the input, every possible combination of prefix/m with m between mask and maxmask was present:
192.0.2.0/30-32
192.0.2.4/31
192.0.2.6/31
-a3 combines the prefixes of the input in a slightly different notation for exact match with the input. In addition to -a2 also combines complete "sets" from the closest less-specific prefix/mask's perspective (which wasn't part of the input). prefix/mask,/mask1-mask2 states, all possible prefix/m with m between mask1 and mask2 had been present in the input. In some cases the result is not optimal or other "solutions" are possible:
192.0.2.0/29,/31-31
192.0.2.0/30
192.0.2.0/30,/32-32
-a4 is suitable as input for Juniper's route-filter(-list) with upto (prefix/mask1-mask2 => route-filter prefix/mask1 upto /mask2) or prefix-length-range (prefix/mask,/mask1-mask2 => route-filter prefix/mask prefix-length-range /mask1-/mask2) to exactly match input prefixes. In some cases the result is not optimal or other "solutions" are possible.
192.0.2.0/30-32
192.0.2.4/30,/31-31
-c adds some statistics at the top of the output, prefixed with #
# Input v4: 9 unique prefixes
# Input v6: 0 unique prefixes
# aggr1 mode
# Output v4: 1 unique prefixes
# Output v6: 0 unique prefixes
192.0.2.0/29
-v complains about non-parsable input, -v -v also complains about unaligned input.
-h gives a short help.
PS1: And do not use this in your data structure lessons ... there are better data structures for prefixes (radix tree / trie) than an unbalanced tree with all intermediate nodes created ... quick hack and so "easy" to understand aggregation, if you draw it on a piece of paper ... and still significant faster than most other aggregate tools.
PS2: Oh, yes, gcc knows 128bit integers, which would allow easier code writing ... but other compilers like Sun Studio don't ...
PS3: This is somewhat old code ;-)
Please report bugs at https://github.com/FvDxxx/pfxaggr/issues
Markus Weber [email protected]