Eclipse Community Forums: OM2M » why do we need mn cse and in cse?

Home » Eclipse Projects » OM2M » why do we need mn cse and in cse?(why do we need mn cse and in cse in om2m and what are their differences?)

Show: Today's Messages :: Show Polls :: Message Navigator

why do we need mn cse and in cse? [message #1738948]

Tue, 26 July 2016 09:28

Son Duong

Messages: 60
Registered: June 2016

Member

Hi all,

maybe this could be a ridiculous question but can anyone explain me why do we need both mn cse and in cse in OM2M and what are their differences? It seems that I can deploy everything including devices, gateway....all in one of them, so I need only either mn cse or in cse running. Am I right?

Thank you in advanced!

Report message to a moderator

Re: why do we need mn cse and in cse? [message #1738956 is a reply to message #1738948]

Tue, 26 July 2016 10:03

Mahdi Ben Alaya

Messages: 229
Registered: November 2013

Senior Member

Hello,

IN: Infrastructure Node, can be seen as a centralized server deployed on the cloud.
MN : Middle Node, can be sees as a gateway deployed locally.

MNs has the possibility to register to IN and so enable one to obtain a highly distributed architecture covering various use cases.

You can consider an IoT architecture including only one IN if your prefer a centralized architecture, however this use case is not interesting because it requires that all your devices and applications are capable to reach the IN which is not the case for small sensors. The MN can act as a proxy to connect local constrained devices to the IN.

You can also consider an IoT architecture including only one MN deployed locally if you don't need a centralized server, however this is not so recommended because it requires the MN to answer to all coming requests. In general, the MN is a constrained device which may represent a single point of failure in this case. Imagine a use case where your sensors are not deployed on the place which requires more than one MN. In this case, the IN will be the intermediate between all existing MNs.

For me, it is better to have a distributed IoT architecture including one IN and several MNs deployed in several places. The IN will be in charge to re-targeting requests from a Node to another in a seamless way. Applications can be deployed on the IN or on the MNs according to your needs. Each node can host customized services like cool dashboard in the IN, or dedicated interworking proxies in the MNs to connect local vendor-specific devices.

Hope this can help
Regards
Mahdi

Report message to a moderator

Re: why do we need mn cse and in cse? [message #1738959 is a reply to message #1738956]

Tue, 26 July 2016 10:24

Trung Luu

Messages: 32
Registered: June 2016

Member

Dear Mahdi,
I'm not quite clear about the statement: " In general, the MN is a constrained device which may represent a single point of failure in this case. Imagine a use case where your sensors are not deployed on the place which requires more than one MN. In this case, the IN will be the intermediate between all existing MNs.". Could you please explain it more precise for me to understand? Thank you in advanced!

Regards,
Trung

Report message to a moderator

Re: why do we need mn cse and in cse? [message #1738960 is a reply to message #1738959]

Tue, 26 July 2016 10:40

Mahdi Ben Alaya

Messages: 229
Registered: November 2013

Senior Member

Let's suppose that you have an MN-CSE deployed on a Raspberry pie in your home. The MN will not be able to answer a big number of requests due to the gateway constraints in terms of CPU, RAM, etc. So such deployment will not scale. In this case, it is better to make the MN pushing some relevant data to the IN (with high performance deployed on the cloud) in such a way the IN can handle directly the received requests without the need to trigger the MN for each request.

[Updated on: Tue, 26 July 2016 10:41]

Report message to a moderator