Depends on your layout, but you could also bond them together in failover mode. Then no OSPF timeout hit.
On Mon, Sep 4, 2017 at 11:52 AM, Kurt Fankhauser <[email protected]> wrote: > OSPF cost for backup route definitely needs to be higher... you can > standardize all you want on some numbering system but as your network grows > and you need to push some traffic in some directions and not others you > will need the flexibility to manipulate path costs quite a bit. > > On Sun, Sep 3, 2017 at 4:59 PM, Faisal Imtiaz <[email protected]> > wrote: > >> Some more food for thought... >> >> >> We are finding that setting up /configuring the OSPF links as PTP tends >> to provide faster failover convergence >> which becomes even more useful when it can be combined with bfd >> >> In regards to the weights, it's more of a 'six or half a dozen of >> another' what values you use to affect the change will be determined by >> what exactly you are trying to achieve and 'flow' of traffic on your OSPF >> network.. >> >> Regards. >> >> Faisal Imtiaz >> Snappy Internet & Telecom >> 7266 SW 48 Street >> Miami, FL 33155 >> Tel: 305 663 5518 x 232 <(305)%20663-5518> >> >> Help-desk: (305)663-5518 <(305)%20663-5518> Option 2 or Email: >> [email protected] >> >> ------------------------------ >> >> *From: *"Steve Jones" <[email protected]> >> *To: *[email protected] >> *Sent: *Sunday, September 3, 2017 4:47:03 PM >> *Subject: *Re: [AFMUG] easy backup link failover >> >> I was thinking about that, 10 doesn't give much room for manipulation >> >> On Sat, Sep 2, 2017 at 9:26 PM, George Skorup <[email protected]> >> wrote: >> >>> That's typically what I do, just make the parallel backup path one >>> higher at both ends. >>> >>> But I'll tell you this right now, consider a larger scale for your >>> interface costs. As your OSPF domain grows into more complex rings or more >>> of a mesh, shit will start to get complicated and you'll wish you had more >>> granularity. What I'm moving to is interface cost based on link bandwidth. >>> Kinda like Cisco's auto-cost, but not auto because MikroTik is stupid. >>> Anyway.. take 100,000 ÷ link bw in Mbps. So 1G=100. An AF24 around 770Mbps >>> would be a cost of about 130. A 360Mbps SAF link would be about 277. Etc, >>> etc. Lots of granularity for tweaking traffic flow. >>> >>> >>> On 9/2/2017 4:08 PM, Steve Jones wrote: >>> >>>> we are replacing two links, currently cheap 5ghz (one epmp ptp and one >>>> ubnt nanobridge) with mimosa 11ghz, we dont need that much bandwidth right >>>> now so im leaving the old links in parallel. >>>> >>>> I just put the path cost on the interface for the 5ghz at 11 and left >>>> the 11ghz at 10. it seems to serve this purpose. but the other links in the >>>> redundancy will see that extra 1 in path cost on failover, not so awful a >>>> deal since it will drop capacity by 90 percent, but would i have been >>>> better to leave the 5ghz at 10 and drop the 11ghz to 5? >>>> >>>> >>> >> >
