Connector Type of Interfaces

[sw8899]

NetBox version 3.2.1

Is there a reason that it is possible to define a Connector Type (e.g. 8P8C, LC, MPO) on front and rear ports but not on interfaces?
Am I misunderstanding a concept or is it just missing on the interfaces?

[mtinberg]

I imagine it's usually less relevant for Interfaces, which already have a type that implies the connector in most cases, eg. a 10GBase-T interface is not going to be LC, but an SFP+ probably is (unless its DA ;-). If you have a need to track something that isn't in the primary data model, you can custom fields to your heart's content to track whatever you need. If enough people report something missing that maybe should be in the built-in models then it can be added and you can plan a migration to it. — Mark Tinberg ***@***.***> Division of Information Technology-Network Services University of Wisconsin-Madison

…

________________________________ From: sw8899 ***@***.***> Sent: Wednesday, September 14, 2022 10:49 AM To: netbox-community/netbox ***@***.***> Cc: Subscribed ***@***.***> Subject: [netbox-community/netbox] Connector Type of Interfaces (Discussion #10364) NetBox version 3.2.1 Is there a reason that it is possible to define a Connector Type (e.g. 8P8C, LC, MPO) on front and rear ports but not on interfaces? Am I misunderstanding a concept or is it just missing on the interfaces? — Reply to this email directly, view it on GitHub<#10364>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAS7UM2RJ2NP73WDTSPH7GDV6HXXTANCNFSM6AAAAAAQMRVDHY>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[DanSheps]

I don't thinik we want to get too far into the weeds on this one, otherwise we start modelling optics themselves which I am not sure we want to do.

[sw8899]

ok, maybe I got lost in translation. I'll try to repeat in my words what I think I understand:
In order to capture connector types on optical interfaces, the data model would have to be converted into a complex model, and that is something the maintainers of Netbox do not think is worth it.
Right?

[jcollie]

Two ways that kinda work around this:

1. Add a custom field to the interface that has the connector type. Not quite as good as a "real" field but should do the trick.
2. Add an inventory item that has the specific SFP, SFP+, or QSFP model number. This should implicitly specify the connector type as presumably the manufacturer of the SFP uses different model numbers for different connectors.

[sw8899]

thx for the input

1. ok
2. Since it is not possible to define inventory item types this would mean that you start over with the definition for every interface again?

[DanSheps]

I believe InventoryItemTypes were proposed at one point, not sure where that went.

But yes, you would need to define it for every interface.

[sw8899]

I understand that there are workarounds. But the main problem remains. They don't show up where you need them: cable trace, cables
My current problem is. The IT/Network departments do all the planning.
Then we (cable monkey department) go into netbox and see all the planned cables.
Then we have to order and patch them.
I get the information Cat6a or OM4 and length and color and label from the cables list in Netbox.
But the most important property of the cable, the connector is missing.
From my point of view I can not understand how anybody can life without this information ;-)

[shatt79]

Just have the networking team standardize on singlemode/LC across the board and you never need to worry or care about it. There once was a cost benefit to using MM. That doesn't exist anymore.

There's literally nothing outside of GBICs or X2s that use anything but LC. GBICS and X2s have been obsolete for a decade.

[sw8899]

I'm not in the position of giving the engineering teams advice.
I see hundreds of QSFP+ 40GBASE-SR4 MTP (e.g. in HP Synergy)

[jcollie]

There once was a cost benefit to using MM. That doesn't exist anymore.

The cost of using existing multimode (where possible) will always be cheaper than installing new singlemode. There are those of us managing existing plants that are decades old. Plus there's all of the existing singlemode that was installed using SC or ST connectors. Telling people to spend millions on new fiber infrastructure to fix a software problem seems a bit out of touch.

[sw8899]

from FSdotCOM: QSFP-40G
MM MTP SR: $ 39.00
SM LC LR: $ 309.00 (cheapest SM) (Does SM SR exist?)

[candlerb]

Does SM SR exist?

No: SR = Short Range = Multimode, pretty much by definition.

Multimode is inherently short range: the different propagation modes interfere with each other. Single-mode doesn't have this problem: its range is limited mainly by optical attenuation.

Both sides are right here. Using existing multimode plant is clearly cheaper than deploying new single-mode, but all new deployments should be single-mode. Even if the optics are a little more expensive, the single-mode fibre itself is cheaper and will accommodate any future bandwidth requirements.