search

kytos-ng / pathfinder

Kytos main path finder Network Application (NApp)
https://kytos-ng.github.io/api/pathfinder.html
MIT License
0 stars 7 forks source link

DISCUSSION: New `spf_attributes` #60

Open Ktmi opened 11 months ago

Ktmi commented 11 months ago

This issue is to discuss new possible spf_attributes to be used for determining the total weight of a path.

One suggestion is the total number of evcs being used by a path. This attribute on its own wouldn't be very good, but if #59 were to be implemented in the future, it would make a great tie breaker. However if we allow unidirectional evcs as suggested with kytos-ng/mef_eline#362, this feature could potentially not work as intended, as it would be harder to account for the direction of the traffic.

viniarck commented 11 months ago

Maybe a unidirectional EVC, if we support it, it could weigh 0.5, whereas a bidirectional EVC would weigh 1 per edge?

viniarck commented 11 months ago

@Ktmi the example you gave on issue #59 looks coherent to me.

I'll recommend that you also confirm with @jab1982 and @italovalcy one production-like example of how they'd use, then we can also map an equivalent one to include in our e2e tests. Other than that, the feature seems clear to me once you also document the proposed changes after confirming the use case is indeed how we're picturing it so far.

viniarck commented 11 months ago

Maybe a unidirectional EVC, if we support it, it could weigh 0.5, whereas a bidirectional EVC would weigh 1 per edge?

I mean, although we could end up in a case where a link would only have many unidirectional A -> Z EVCs, in practice the link utilization would go up anyway, so it could further also include utilization as a mandatory primary_constraints, still, utilization isn't being populated by a third party or another NApp yet, but I could see this potentially working to achieve a load balance distribution.