Server Issues

Missing data 2.00am to 2.30amm. Have the server's been playing up again?

Also missing same.

Same here, guess GE still having issues there

The server issues excuse is a bit rubbish, really. If they are having server issues, then the data should be late, not missing. It's not hard these days to implement a cloud solution that means that data gets queued up. Azure EventHubs, Amazon Kinesis or GCP Pub/Sub - they are all specifically designed for the purpose. The fact that GE's infrastructure loses data so often should be ringing alarm bells somewhere.

I mean, this stuff is my day job, and it's really not that hard.

The inverter doesn't seem to store anything locally, so losing WiFi temporarily also misses data, which seems fairly common from my experience so far.

A lot of these problems would be solved if the inverter had a local cache.

MattWhitfield Exactly. Actually losing the data rather than delayed data is a significant problem and as you say, should be ringing alarm bells. Even more important is that the system is used for control - and you can't if their servers are down. Needs sorting urgently I'd suggest.

@LordHippos if losing WiFi meant losing data, fair enough. I mean, they could have a small cache in the inverter to store data in that scenario, but, whatever, it would be at least understandable.

To run a cloud service that completely fails to ingest any data whatsoever when there is an issue is not the same problem...

MattWhitfield

I understand I am just saying that both options would be addressed with a local cache that can store at least some data in case of network connectivity loss, it would allow it to catch up and we'd all have more accurate figures.

LordHippos yeah, I don't disagree - I think a local cache would be useful. But it is hard to scale. If there is a big outage, maybe you still lose data, unless you want to go and retrofit bigger memory chips on each inverter (bear in mind I have no idea how much storage they have right now).

Scaling a service centrally to cope with a lot more queued up data is pretty easy if you have made the right architectural choices on your cloud platform.

Ideally they would do both. If it was up to me, I would work on the server side first.