Ressource consumption when sending data from AWS Greengrass to the AWS Cloud

Question

I'm currently working on a project where we are using AWS Iot Greengrass on an edge device to measure some metrics coming from a machine. We have a Kinesis Analytics application running to detect some anomalies in the data. My problem is that I really don't know how to efficiently send the data to the aws cloud with a mostly low memory foodprint.

I see in the greengrass sdk when I use the IotDataClient (we are using pyhton) every invocation on publish is calling internally the a lambda function. I fear (we need to send at approx 500-1000 datapoints/s) that this might kill the edge device and utilise high cpu consumption.

The other way is to use the StreamManager which might be a better fit for high volume data but I'm not sure if this is a little overkill since it brings a java process to the device. Also I cannot publish to MQTT and I'm tied to AWS Kinesis or other managed services.

Does anyone have the experience with loading this amount of of data via greengrass to the AWS Cloud?

Test it, it's the only way to be sure with your specific setup. — hardillb, Mar 03 '21 at 08:35
but it seems more specific to aws in this case. As far as I know mqtt protocol can easily handle this amount of data — emilio, Mar 03 '21 at 09:06
But it's going to be specific to your choice of edge device, your specific messages and the network connection. — hardillb, Mar 03 '21 at 11:22
true that. I'm just curious if there are any experiences with some kind of setup — emilio, Mar 03 '21 at 13:12
Test, test and retest. Else, failure. See this video for one test that seems relevant https://www.youtube.com/watch?v=TxA1CmZrVyM — Gambit Support, Mar 03 '21 at 14:33

Ressource consumption when sending data from AWS Greengrass to the AWS Cloud

0 Answers0