Re: Comments on DMM Gap Analysis.

On Thu, Sep 27, 2012 at 10:32 AM, jouni korhonen <[email protected]> wrote:
>
> I am inclined to agree with Wes that DMM is hard enough already without
> having clouds arising in the horizon. Cloud computing and such has a
> premise but I have hard time seeing how that would drive distributed
> mobility management. I rather see it as opposite.. shifting the mindset
> again towards centralized data centers.
>

Yes, exactly. So cloud is going in a different direction then DMM,
unfortunately.

> On the other hand if we just think mobility management entities as VMs,
> spawning new functions here and there does not differentiate from what
> we are already trying to figure out - from mobility protocols point of
> view. The difference would be that the possible gateway is not a physical
> box anymore. The herding of VMs is not what we are tasked to do.
>

Here the issue is these VMs doing mobility functions.
We need to develop a different set of goals for DMM implemented in VMs.
What are the issues there? Certainly different than the issues we are
considering now.
Will control plane/data plane separation be the central issue?


I am not saying we should not work on the current DMM direction. We
can work on both.

Regards,

Behcet


> - Jouni
>
> On Sep 26, 2012, at 11:08 PM, Behcet Sarikaya wrote:
>
>> On Wed, Sep 26, 2012 at 2:48 PM, Wesley Eddy <[email protected]> wrote:
>>> On 9/26/2012 3:40 PM, Behcet Sarikaya wrote:
>>>> I am talking about 3GPP entities like Serving Gateway and PDN Gateway
>>>> being placed in the cloud.
>>>
>>>
>>> Sure, so they run on a VM somewhere ... what's the technical
>>> challenge that this poses?  To the mobile node and rest of
>>> the DMM infrastructure, it should not matter whether these
>>> are physical or virtualized.
>>
>> DMM is based on the premise that mobility entities come closer to the
>> UE and UE can access a different one as it moves. Central anchoring in
>> PMIPv6 or MIPv6 is no longer there.
>>
>> With the cloud, two mobility entities may now become IPv6 neighbors in
>> the cloud. Trying to define signaling between such entities maybe no
>> longer a big issue.
>>
>> With cloud, DMM's premise somewhat disappears. We need to find a new premise.
>>
>> Regards,
>>
>> Behcet
>>
>>
>>>
>>> --
>>> Wes Eddy
>>> MTI Systems
>