Sure you can. Run something like quagga on the server for OSPF, configure your
1g links as /31s. The key is to attach your service to a loop back IP on their
server and advertise that loop back using quagga to your router. I'd also make
each of the 1g links a point to point ospf type.
nkering
with QoS is cheaper than not, I agree, go with QoS.
--Andrey
On Tue, Nov 15, 2016 at 2:44 AM, wrote:
> > Of Andrey Khomyakov
> > Sent: Monday, November 14, 2016 4:56 PM
> >
> > OP explicitly stated that there is not congestion. I think we can all
> agree tha
"My opinion on QoS for networks with low bandwidth is to always implement
it." Most definitely. Scheduled drops are always better than tail drops.
As for microbursts, I can't say I'm very knowledgeable about what that is.
I'm guessing that we are talking about extremely brief periods when egress
ra
ce or
shape (I know this is juniper list, but I don't know junos equivalents)
statements to below the line rate.
--Andrey
On Mon, Nov 14, 2016 at 12:43 AM, wrote:
> > Of Andrey Khomyakov
> > Sent: Thursday, November 10, 2016 9:28 PM
> >
> > is deployed. As usual, YMMV,
Sure, you could mitigate some of that with enforcing w/ever at the
boundaries, but hopefully you are starting to see how a simple QoS policy
design on a single node is blowing up into managing hundreds/thousands of
ports and making sure you trust your VoIP telephone, but not the
workstation behind
You could end up in a DoS situation where some rouge source send 1gbps worth of
packets matching your priority class. That's just of top of my head.
Also, depending on what you decide to match on (e.g IP, DSCP marks, ingress
iface, etc) you'll constantly will have to tinker with the policy whenev
My understanding was that if you don't have congestion, QoS is simply
useless for a number of reasons:
1. You spend engineering hours writing/debugging policy that will never get
used (we assuming no congestion)
2. QoS on egress will not go into effect (other than incrementing counters)
because th
7 matches
Mail list logo
