Taking a quick look at this it doesn't seem to offer much more over AWS. Device management and remote access might be nice. But its been my experience in B2B land that anything like this gets locked down by internal IT depts.
It might be a useful service for B2C devices though.
AWS has an IoT platform that offers roughly the same features. They also have "Device Shadowing" which lets you store up some state that behaves as the device when the connection is lost. https://aws.amazon.com/greengrass/ also looks interesting if you requires some edge computing services for your IoT devices.
I don't actually use either of these platforms currently. I put together an in house platform for the few features I need. Provisioning new device firmware and updating are probably my biggest pains. Yocto is kind of awful as well, but its the best that we have right now.
Giving the device I unique ID, registering it somewhere. Mount the image and inject configuration information. Then you need to build some production worker friendly appliances to actually program your devices.
This is great and all except I had registered the company I work for to be on the private beta, and have not heard anything for weeks. There's not even a place or number or email to contact, it's just a Google Form. I even signed up for their Brillo/Weave program (now Android Things), and also heard nothing back. You either have to have some internal connection to get any attention or just wait for their service to go into production, which is God knows when.
It is more than one year that I work with IoT and honestly I don't have understood the google offert.
I had the pleasure to work with IoT and the Lora protocol writing and managing a Lora network server, the piece of software that takes Lora messages send through radio and decode them from radio -> udp -> MQTT (encrypted) -> MQTT (clear)
[shameless plug, I packed my experience in an on-premises service: http://loranetworkserver.com/
]
The difficult part I believe is this one, after you got the MQTT message in clear what you do with it is quite simple, straightforward and overall a solved problem (unless you are not getting millions of messages per seconds, but very few businesses have such load).
From what I understood from the google page they are solving the simpler problem of getting the data from an MQTT message and save it to disk, and maybe use later for data mining.
It is weird because I wouldn't accept to have my data saved in someone else disk, at the bare minimum I would require to have the data duplicated in some standard (pg, MySQL, mongo) db (even a db managed by AWS or Google, but a standard one that I can move at will).
It doesn't seems the most valuable piece of architecture that they could develop.
The real valueble piece would be probably something up the stack, maybe similar to the great offering of thingworx, but better.
But that would mean competing with partners , so it has its own issues.
Maybe by focusing on something like authentication , and offering Google branded security , they could both achieve some lock-in in and not scare partners ?
My point of view is quite local, however I don't see any of the company I work with having their data only on Google or AWS without a very clear way to get it out.
Honestly I see IoT a lot like Machine Learning, everybody is talking about it, everybody want to be part of the game, but there is not a clear solution.
Businesses have very specific use cases that require some, not trivial, understanding of the technology, to be fulfilled in full.
Even more to actually use the data gathered.
For the foreseeable future I see the need to a quite senior figure to enable IoT in businesses.
Once a senior figure has such data in clear, a way to mine value out of that data I see at least unlikely that she is going to push all of that only on google only to fetch then back once you need to do your visualization or analysis.
Maybe if you are completely sold to the google ecosystem, it may make sense, but still...
Honest question, I don't understand these "IoT servers". Aren't they just standard servers that accept agreed protocol? E.g. HTTP or MQTT? Isn't this something you can do yourself with $5 Digital Ocean?
16 comments
[ 1.3 ms ] story [ 68.1 ms ] threadIt might be a useful service for B2C devices though.
I don't actually use either of these platforms currently. I put together an in house platform for the few features I need. Provisioning new device firmware and updating are probably my biggest pains. Yocto is kind of awful as well, but its the best that we have right now.
Works pretty well, and the Business Intelligence backend is pretty powerful. Not sure what Google's version of that is here.
I had the pleasure to work with IoT and the Lora protocol writing and managing a Lora network server, the piece of software that takes Lora messages send through radio and decode them from radio -> udp -> MQTT (encrypted) -> MQTT (clear) [shameless plug, I packed my experience in an on-premises service: http://loranetworkserver.com/ ]
The difficult part I believe is this one, after you got the MQTT message in clear what you do with it is quite simple, straightforward and overall a solved problem (unless you are not getting millions of messages per seconds, but very few businesses have such load).
From what I understood from the google page they are solving the simpler problem of getting the data from an MQTT message and save it to disk, and maybe use later for data mining. It is weird because I wouldn't accept to have my data saved in someone else disk, at the bare minimum I would require to have the data duplicated in some standard (pg, MySQL, mongo) db (even a db managed by AWS or Google, but a standard one that I can move at will).
It doesn't seems the most valuable piece of architecture that they could develop.
But that would mean competing with partners , so it has its own issues.
Maybe by focusing on something like authentication , and offering Google branded security , they could both achieve some lock-in in and not scare partners ?
Honestly I see IoT a lot like Machine Learning, everybody is talking about it, everybody want to be part of the game, but there is not a clear solution.
Businesses have very specific use cases that require some, not trivial, understanding of the technology, to be fulfilled in full. Even more to actually use the data gathered.
For the foreseeable future I see the need to a quite senior figure to enable IoT in businesses.
Once a senior figure has such data in clear, a way to mine value out of that data I see at least unlikely that she is going to push all of that only on google only to fetch then back once you need to do your visualization or analysis. Maybe if you are completely sold to the google ecosystem, it may make sense, but still...