Re: OSPF question
Stephen Zedalis (tintype@exis.net)
Sun, 4 May 1997 22:11:37 -0400 (EDT)
The fly in the ointment with your solution is that to align the network to
a /27, you give it an assigned address of 0,32,64 etc. This has the
unsavory result of giving a host address of .0 and the first host on every
portmaster the same address as the corresponding subnet (I said unsavory,
not impossible). I would rather live with Livingston's creative
subnetting. I rather even more, that there would be a option or whatever,
to advertise a single subnet route to the address ONE-LESS than the
assigned address of the first host in the pool. There it should be fairly
easy for most PM-2 configurations to get the pool subnet to one aggregated
route advertisement (and possibly one or two for a PM-3). And it should
be possible to maintain an unsullied network address for reference by
other routers. (broadcast address is moot, as the pool refers to a
non-broadcast medium)
On Sun, 4 May 1997, Curtis Coleman wrote:
>The effect you are looking at is, more graphically:
>
> .1 -------- /32
> .2 \_______ /31
> .3 /
> .4 \
> .5 \______ /30
> .6 /
> .7 /
> .8 \
> .9 \
>.10 \
>.11 \____ /29
>.12 /
>.13 /
>.14 /
>.15 /
>.16 \
>.17 \
>.18 \
>.19 \____ /29
>.20 /
>.21 /
>.22 /
>.23 /
>.24 \
>.25 \______ /30
>.26 /
>.27 /
>.28 \_______ /31
>.29 /
>.30 -------- /32
>
>The routes above are covering 30 addresses, not 32. If you are really
>troubled by all these routes (even though they are persistant), increase
>the pool size to 32 and align it so a single /27 can cover it. It's
>nothing to worry about, I actually think the best-fit algorithm is kinda
>cute.