[Netslices] netslicing/netslices: Potential next steps

[Benoit Claise <[e-mail]>]

Dear all,

Thanks for all who contributed to the netslicing BoF: the document 
authors, the presenters, the participants, and the chairs. During this 
week, next to BoF itself, I've had multiple opportunities to speak about 
netslicing, with some BoF proponents, with the IAB shepherds, and with 
the IESG/IAB on Friday afternoon, etc. This helped me understanding the 
different points of views and clarified some aspects.

It's true that the network slicing has got a specific meaning in the 
context of 5G, but, as Georg (as liaison) stressed: there are no 3GPP 
requirements for the IETF at this point in time. Nevertheless, there are 
parts that the IETF should look at. And there are parts of the network 
slicing concept that are already being worked on at the IETF. ACTN comes 
to mind, but also DETNET, and I heard of VPN+. The individual work in 
WGs should continue and potentially be extended.

Note that I carefully used the "network slicing concept" term in the 
previous sentence. As one important next step, it would nice to focus on 
a common "network slicing" definition, at least its meaning for the IETF.

As mentioned during the BoF, this is not the IETF role to define a 
complete orchestrator, controller, or NMS/OSS. This should be developed 
by an opensource project. However, what would be a practical next step 
is to specify a kind of network slicing template, with the envisioned 
required parameters. From there, we could determine what's missing when 
we try to map it to a specific (IETF) technologies.

Let me make up an (hopefully simple) example: I would like to have a 
network slice with a guaranteed bandwidth b, end-to-end jitter j, packet 
loss p, end-to-end max delay d. Do we even those abilities in the 
underlying technologies?

Regards, Benoit (as the BoF responsible AD)