[JIRA] Commented: (MDA-50) create stage to consolidate mdui:IPHint CIDR blocks
Ian Young (JIRA)
noreply at shibboleth.net
Mon Sep 19 22:01:25 BST 2011
[ https://issues.shibboleth.net/jira/browse/MDA-50?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13567#comment-13567 ]
Ian Young commented on MDA-50:
------------------------------
Oh, credit Brook Schofield for thinking of this one. Also, he came up with this second example:
212.219.247.128/26
212.219.247.192/27
212.219.247.224/28
212.219.247.240/28
Become:
212.219.247.128/25
(two /28s become a /27, the two /27s then become a /26, the two /26s then become a /25...)
Both examples are from live UK federation metadata.
> create stage to consolidate mdui:IPHint CIDR blocks
> ---------------------------------------------------
>
> Key: MDA-50
> URL: https://issues.shibboleth.net/jira/browse/MDA-50
> Project: Metadata Aggregator
> Issue Type: New Feature
> Components: SAML Metadata
> Affects Versions: 0.6
> Reporter: Ian Young
> Assignee: Chad La Joie
>
> CIDR blocks in a federation registry are probably arranged in terms the entity owner understands, for example one for each local subnet. We see this in the UK federation in the schools sector, with many individual ranges.
> It would be a slight optimisation to have a stage which can consolidate these into shorter prefixes. For example, the following ranges:
> 194.80.19.0/26
> 194.80.19.64/26
> Can be replaced by this one:
> 194.80.19.0/25
> I think this is fairly simple in principle, if you first order the list by both prefix length and address. Then, you can repeatedly remove any pair of entries which have the same prefix length and differ only by the last masked bit in their address and replace it with the lower of those addresses but with a prefix shorter by one. Any block which is contained entirely within a block earlier in the list may also be removed.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
More information about the commits
mailing list