[bdNOG] Service level effect in SM-MM fiber combination

Thu Feb 25 00:25:07 BDT 2016

Hi Brian,

Thanks a lot for your thoughtful and informative analysis. We're using
1000BaseLX SFP, I push our team to go with your first recommendation making
it uniform with SM (or MM) along the full path.

BR,
Awal

On Wed, Feb 24, 2016 at 9:23 PM, Brian Candler <b.candler at pobox.com> wrote:

> On 24/02/2016 14:18, Md. Abdul Awal wrote:
>
> We have a probable link to provision with a SM-MM-SM fiber combination
> shown below:
>
>
> 
> Wondering if there is any issue in terms of QoS, data flow or any other
> service related problem in the long run. I'd appreciate your comments on
> that.
>
> I would strongly recommend you don't do this: the amount of light which
> can get from the MM into the much narrower SM is very low, and the majority
> will be reflected back.
>
> What interfaces are you using at each end? A 1000baseLX SFP is capable of
> running over either MM or SM, so if you simply use SM patch cords at both
> ends then it's SM end-to-end, and all the problems go away.
>
> If you are using 100baseFX media convertors then it might work, but media
> convertors are unmanaged and notoriously unreliable. Putting an SFP in a
> managed switch will be a much more robust solution.
>
> Something like the Netgear GS110TP costs about $150 and gives you two SFP
> ports, 8 gigabit copper ports (with PoE!), and is fully manageable (SNMP,
> HTTP, and telnet on port 60000)
>
> Regards,
>
> Brian.
>
> P.S. There are special "mode conditioning patch cords" you can get, but
> these are used for the opposite situation - i.e. where the long link is
> multi-mode and the tails are single-mode. (The idea is to get slightly
> longer reach from existing MM plant)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.bdnog.org/pipermail/nog/attachments/20160225/0c292c5c/attachment.html>