Re: [Nfvrg] [Detnet] [5gangip] Network Slicing - a suggestion that we meet to discuss in Seoul

> On Nov 2, 2016, at 3:40 AM, Pascal Thubert (pthubert) <pthubert@cisco.com> wrote:
> 
> Looks correct, Joe, but then overlays as we traditionally see them do not necessarily provide the degree of isolation that we seem to be after here.

It seems like you're just noting that most Internet overlays don't bother with provisioning...

> 
> And this is where DetNet comes into play. At the extreme of the definition, a deterministic flow will traverse the network is such a fashion that the influence from other flows is not observable on that flow, e.g., no jitter and no congestion loss induced by other flows holding or clogging a link somewhere down in the fabric. Which means that one overlay will experience the same connectivity as if the nodes that compose it were physically meshed with serial wires. 

If you're talking about a flow on an otherwise empty overlay, that is just a fully provisioned overlay. If you're talking about multiple flows within an overlay being isolated this way, it's not and Internet overlay. Maybe an ATM overlay.

But these are still IMO examples of provisioning - the kind we have been dong since the phone network for over 50 years. No need for new marketing terms.

Joe

> 
> Cheers,
> 
> Pascal
> 
> -----Original Message-----
> From: detnet [mailto:detnet-bounces@ietf.org] On Behalf Of Joe Touch
> Sent: mardi 1 novembre 2016 23:35
> To: Stewart Bryant <stewart.bryant@gmail.com>; 5gangip@ietf.org; detnet@ietf.org; nfvrg@irtf.org; draft-galis-anima-autonomic-slice-networking@ietf.org; draft-vonhugo-5gangip-ip-issues@ietf.org; draft-xuan-dmm-multicast-mobility-slicing@ietf.org; draft-ietf-teas-actn-framework@ietf.org
> Cc: Dongjie (Jimmy) <jie.dong@huawei.com>; Mach Chen <mach.chen@huawei.com>
> Subject: Re: [Detnet] [5gangip] Network Slicing - a suggestion that we meet to discuss in Seoul
> 
> 
> 
>> On 11/1/2016 12:39 PM, Stewart Bryant wrote:
>> Joe, we possibly have different ideas as to what network slicing is.
>> 
>> My assumption is that it is a way that we can create multiple 
>> virtually networks over a common base network with a higher degree of 
>> isolation between VPNs and much higher performance guarantees than we 
>> currently achieve.
>> 
>> If we need a new name, we need a new name, but the concept does not 
>> seem to align with the definition you hints at.
> 
> Right. That's because you're describing an overlay.
> 
> Overlays have varying degrees of isolation and performance (and isolation and performance assurance), and always have.
> 
> There's no new term needed for that.
> 
> Joe
> 
> _______________________________________________
> detnet mailing list
> detnet@ietf.org
> https://www.ietf.org/mailman/listinfo/detnet