+2 to Thiago?s opinion.
@Vandita
I have tested both the CoAP and CoAP + TCP for implementing file transfer
feature.
and..
Case 1. with CoAP + TCP (with SECURED=1), it did not exceed 2 min when
transferring the 50MB.
Case 2. with CoAP (with SECURED=1), it did not exceed the 5 min when
transferring the 50MB.
I have to admit that those results are out of date, meaning I have tested case
2 more then 4 months ago.
I will test those 2 cases again and share you the result asap.
However, I think the performance depends not only on file size but also the HW
capabilities.
What is the HW capabilities that you?re using?
Thank you.
Jay.
> 2017. 3. 2. ?? 9:30, Thiago Macieira <thiago.macieira at intel.com> ??:
>
> On quarta-feira, 1 de mar?o de 2017 09:54:55 PST Venuturapalli, Vandita wrote:
>> The problem here is since the performance depends largely on the size of the
>> file, a file size of 68KB takes less than 3 seconds whereas file of size
>> 7MB takes 3.14 minutes. Did you run into the same problem on transferring a
>> file size of 50MB?
>
> 7 MB is already much bigger than what can be expected from constrained
> devices. The entire device image is probably under 2 MB, probably much lower.
>
> For non-constrained devices, I'd say you should negotiate a non-OCF data
> transfer, like one based on CoAP+TCP or HTTP. Just be really careful about
> the
> security implications, as the extra open port could be accessed by an
> attacker
> pretending to be the device you communicated with. I'd make this transfer
> encrypted and require a nonce in the URL that was transmitted over OCF.
>
> --
> Thiago Macieira - thiago.macieira (AT) intel.com
> Software Architect - Intel Open Source Technology Center
>
