auto-cost reference-bandwidth

Ondrej Zajicek santiago at crfreenet.org
Tue Jun 13 14:09:31 CEST 2017


On Wed, Jun 07, 2017 at 06:32:59PM +0200, Markus Wigge wrote:
> Hello,
> 
> according to a chat I recently had with Elrond I put my question on the
> mailing list:
> 
> Is there an option for the OSPF protocol in bird to calculate the
> interface cost based on the bandwidth like cisco or quagga or juniper or
> fortigate or ... do?

Hello

Short answer: There is not.

Long answer: There are two arguments against it:

First, it is something that IMHO does not make sense. Usually, one would
want to maximize bandwidth, so e.g. 3 hops of 100 Gbps are better than
one hop of 40 Gbps. Generally, i would suggest to split links to classes,
then assign cost to each class in such a way that paths consist of links
of preferred classes are better than even a single hop path of less
preferred class. Even if one would like to have formula from bandwidth to
cost, it would make sense to have something superlinear to inverted
bandwidth.

But there are many possible cost policies (with multiple parameters), so
it makes sense to do that outside of BIRD and i don't see a reason to
hard-wire one such policy directly into BIRD, esp. when i don't see it as
a good policy compared to alternative ones.

Second, it is nontrival thing to do. There is AFAIK no simple syscall
'get_bandwidth', you have to use multiple ways based on OS and interface
type. In some cases it is even dynamic (e.g. on WiFi). So generally the
utility not worth the effort.

-- 
Elen sila lumenn' omentielvo

Ondrej 'Santiago' Zajicek (email: santiago at crfreenet.org)
OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net)
"To err is human -- to blame it on a computer is even more so."


More information about the Bird-users mailing list