The MPLS WG Archive

["Cheenu Srinivasan" <cheenu@alphion.com>]

Igor Lasic wrote:
>
>   In networking Access Lists are a general and well understood
>   concept. Access lists give one ability to collect matching 
> information
>   and then "bind" it to a entity.
> 
>   For example:
> 
>   acl 1
>       10 ip dest address 1.1.0.0 255.255.0.0
>       20 ip src address 2.1.0.0 255.255.0.0
>       30 allow all
> 
>   Where 10, 20 and 30 are priorities for matching criteria. 
> 
>   Then one just binds 1 (the acl index) to entity (in this 
> case it could
>   be an outsegment entry.) Using the mplsFTNMapTable (I don't 
> think prev
>   and curr would be neccessary.)

We introduced the linked-list structure in this MIB specifically
to overcome a problem with what you describe - everytime an
entry needs to be inserted into an existing list the ids/"priorities"
of existing items may need to be changed. Either that or one
has to think ahead and leave big enough "holes" in the ids to
allow insertion at a later point (which over time still does not
avoid the above issue).

Implementation wise, one could
continue to internally assign IDs/priorities/whatever and renumber
behind the scenes and save the administrator this pain.

>   Secondly, in the mplsFTNEntry ranges used for matching ip addresses
>   are very uncommon way of matching ip routes. In general 
> people use ip
>   address/prefix concept.

Ranges are a superset of prefixes; implementors who choose 
to restrict their implementation to prefixes only can still do so
by restricting the type of ranges thay accept.

Cheenu