Hi Tony,
please see inline (##PP):
On 17/07/2026 17:09, Tony Li wrote:
Hi Peter,
if you read the DCM draft carefully, you will find out that what you
put below is maybe funny, but not more than that.
It’s also accurate. It’s very clear where the intellectual inspiration
for DCM came from. Yes, you’ve changed the details, but you’re
solving the same problem. You didn’t like our solution, so you came
up with a worse one all on your own. And then you berate someone else
for doing what you’ve done.
##PP
I see thinks differently, but let's not argue about that, does not lead
us anywhere.
Here are the significant differences between the two:
1. TTE is based on a backup paths, DCM calculates its own offload paths
More accurately, DCM offloads traffic onto an oscillating Flex Algo
topology. You’ve exactly recreated the BBN SPREAD oscillator from the
Arpanet, one of the classic mistakes in networking
(https://www.cs.princeton.edu/~jrex/teaching/spring2005/reading/khanna89.pdf).
SPF in an unstable topology is going to continually generate
transient micro-loops.
##PP
The problem described in that paper is a result of the entire traffic
shifting between links A and B. That is not what DC does. DCM always
keeps the traffic on the link A and moves part of it on B. It also
monitors B's utilization and stops adding more before B becomes
congested. DCM is also done for local links only, which is different to
what is described in that paper, where all routers react. So we are far
from conditions described in that paper.
2. LFA/TI-LFA does not take congestion into account, DCM offload path
is calculated to be congestion free
DCM’s calculation is at best optimistic. Traffic is fractal.
Congestion happens in spikes. If you do detect congestion, DCM has to
change the Flex Algo topology, thereby thrashing the IGP, and the
entire network has to SPF for that topology. This turns a congestion
event into an IGP churn and micro-loop. This is not better, this adds
to control plane instability. Then after shifting traffic, the link
is no longer congested, so is reinstated to the offload topology and
the cycle repeats.
##PP
We do not react to spikes. And we do incremental offload. All to avoid
what you described.
2. TTE does not support multiple congested links in the network at
the same time. Using multiple LFA/TI-LFA paths simultaneously on
different nodes can easily lead to loops.
One can avoid this via SSPF.
##PP
if your protection is via local LFA, which does not use any extra
labels, SSPF is not going to help you, because your backup traffic is
using the native SID.
3. TTE is proposed to be used for a short duration of the time and
then the TE is expected to recompute. DCM is meant to be used for
long time and does not need any TE action.
DCM keeps the network thrashing indefinitely.
##PP
I don't think so.
4. As a result of (3), DCM offload path is fully protected by
LFA/TI-LFA, TTE paths are note protected.
TTE paths are load balanced, until paths are recomputed, so enjoy full
protection. DCM paths are necessarily transient.
##PP
if you put the traffic on a backup path, it has no protection.
thanks,
Peter
Tony
_______________________________________________
rtgwg mailing list -- [email protected]
To unsubscribe send an email to [email protected]
_______________________________________________
Lsr mailing list -- [email protected]
To unsubscribe send an email to [email protected]
