ST v2 - what can the hub do w/o power and only local LAN access?

Being reliant on internet/cloud access for all functionality has always been a non-starter for my home automation - so the V2 design that provides on-board processing was a key improvement. After all, if my internet is down (say Comcast or a backbone carrier is out), why should that prevent my home automation system from turning on the garden lights at dusk.

So, if the V2 Hub sits on the in-house LAN, my various zigbee/zwave (and future bluetooth) devices within range and I have electrical power - what works and what doesnt?

Iā€™d read on some review sites that what can run ā€˜on hubā€™ (without cloud access) is highly restricted. Is that a per-program flag I, the owner/manager, can set?

Can the phone interact with the hub if the internet access is down, but Iā€™m in the house? It should, but can it.

If Iā€™m in the house, my phone and the hub are on the house LAN, they can communicate over wifi with each other. They should be able to discover each others LAN side IP addresses normally. So I should be able to use the phone to turn off/on my outside light or adjust my thermostat, or look at the feed comming from my front door camera, even without internet access. But can I?

Then if the power goes out and the hub runs on its 4 AA batteries, I can understand that in this case the very limited power would be an issue - and restrict to only running progamming related to smoke and safety sensors and alerts - so the ability to turn on and off outdoor lighthing would o away. For that case, Iā€™d just buy a larger UPS system to power the hub so it doesnā€™t even know the power is out.

Right now almost nothing runs locally.

One smart app, smart lights, and only for specific device types. Even the official integrations with other lan devices like the Phillips hue bridge wonā€™t run when the Internet is out. Basically your hub wants to go to your cloud account in order to see if itā€™s authorized to talk to those devices. Every single time.

And you have no UI over the lan. If you donā€™t have an Internet connection to your hub, your phone app will not talk to it. thatā€™s certainly possible (Staples connect does it), itā€™s just not how SmartThings does it.

Smartthings is still very much a cloud-based architecture. They are moving in the direction of more local processing, but theyā€™ve taken maybe the first of 25 steps to get there.

For more detailed information, see the following topic:

1 Like

I was very disappointed when I learned that things donā€™t all run locally. I waited for v2 before buying as I thought it was the version for everything to run without needing an internet connection. I didnā€™t do my research here first, unfortunately. When my internet connection went down, very basic things like turning on a light based on motion quit working.

I think we were all surprised by how little local processing V two delivered. Including some of the SmartThings staff.

in previous conversations for several months, I think most of us expected it to be much more like Staples connect or Homeseer or vera or for that matter harmony, where you needed the Internet to set up your initial account, but if the Internet wasnā€™t available most things would continue to run locally. In the case of staples connect that also includes the phone app as long as your local Wi-Fi is up.

Instead, the architecture pushes out anything that will run locally to all customer accounts. That meant they had to really limit what it was. So it came down to The one official app, SmartLights. And perhaps now a few things from Smart Home monitor with the most recent firmware update, but we havenā€™t gotten those details yet.

They said they hope to eventually give us the ability to download custom smartapps and device types to our own individual hubs in the future, but they just donā€™t have a way to do that yet.

2 Likes

Thanks for the feedback. That is valuable to know. It effectively means that Smart Things V2 is not even an option.
If I have to have internet connectivity to make the automation work at all, the hub is a non-starter.
This is the same idiocy as Iā€™ve had with Nest (maybe now that their a google property but not sure) - where if there is any issue with their cloud service, I cannot adjust my thermostat from inside the house with the app.

The whole point of automation is so I DONT have to get up and go over there to manually change the setting - and if everything I need is inside the house, on the same LAN, and we all have power, the it must work.

So Iā€™ll go back to see what Vera does with its new hub, if that much delayed and overdue hub (which does do local processing and is supposedly adding zigbee to its existing zwave radio as well as a few other backup options) does, once it comes out (now almost a year after its inital announcment).

Iā€™m sure there is value in the usage data and monetizing that is part of the companyā€™s business model, but you can still get that useage data with usage reports that phone home occasionally - you donā€™t have to rely on routing everything thorugh a cloud service; thats the wrong way to do automation and security. Its why the Nest is a problem, and why DropCamā€™s are limited in funcionality.

Yeah I was really surprised at the end result considering all the previous discussions and all the anticipation.

I did get the pre-order offer to jump in and actually let it go, because I wasnā€™t sure about how things would work ā€“ and I wanted to make sure I wouldnā€™t spend a ton of time migrating my stuff over to then find out it didnā€™t make a difference.

Sure enough, Iā€™m still on the first hub and donā€™t feel like Iā€™m missing anything. Hopefully soon we will see compelling reasons to move to the ā€œnewā€ hub, but right now it doesnā€™t seem worth the trouble.

2 Likes

Iā€™m gonna throw a positive spin on this.

SmartThings has been working like crazy in the background over the last few months to add support, even if itā€™s basic support to lots of devices. Iā€™ve learned this simply by trial and error and others postings.

I currently have 38 of my devices that run locally. I have a total 67 physical devices. I have some other future device swaps and changes and updates to get even more devices local.

My ā€œproblemā€ is that only 16 of my apps run locally and that is because I have a lot of non local devices. So as I resolve local device issues the local apps will increase.

There is always gonna be apps/device types that the community develops that are way better than SmartThings so itā€™s about weighing needs.

What makes a device a ā€œlocal deviceā€. Iā€™d think everything in the house is local.

1 Like

SmartThings maintains your account information in its cloud on its servers. For most things, the logic that decides exactly what messages will be sent to the devices in your house is processed in that cloud. So even if you Say have a Motion sensor that you want to trigger a light coming on, it would work like this:

The motion sensor detects motion

The motion sensor sends the message to its coordinator which is the hub at your house

The hub sends that message to the cloud to figure out what your account wants to do when that motion sensor detects activity

A program for your account runs in the cloud and decides that a message should be sent to turn the light on.

The cloud sends that instruction back to the hub at your house

The hub at your house sends the light the message to turn on.

The two physical devices, the motion sensor and the light, are indeed ā€œlocalā€ to your house. But the programs that decide what to do with that are running on a server in Minnesota or Chicago or wherever. In the SmartThings cloud.

This is how everything worked for the V1 hub.

With the V2 hub, a few, and at the present that is very few, of these evaluation programs can run on a little computer inside your hub at your house. So that is ā€œlocal processing.ā€ As opposed to ā€œcloud processing.ā€

The processing programs allow many devices to work together

The processing is what makes it possible for smart things to use, for example, a Zigbee motion sensor to trigger a Z wave light. Or vice a versa. Because there is a program in between that can reformat the messages for each device so that they donā€™t have to talk directly to each other. That part of the architecture is generally good idea. It gives you a lot more flexibility in device selection.

Then the question is where should those programs run. If they run in the cloud, we may be able to run much more complex programs much more cost-effectively, but you do introduce Potential instability and lag.

Just as an example, almost all voice processing these days is done in the cloud because this is very complex programming. So whether itā€™s hey Google, hey Siri, Amazon echo, or your smart TV with a voice remote, the odds are really high that that processing get sent to a server in some other city. Thatā€™s justified for voice processing, again because of the complexity involved.

Whether itā€™s justified for a motion sensor turning on a light switch, however, is a whole different question. :sunglasses:

2 Likes

Now that Iā€™ve learned where to see what devices and SmartApps are local, Iā€™m a little less concerned than I was previously. Most of my (temporarily) small setup is local. The exception seems to be my Cree bulbs.

The person had used the term ā€œlocal deviceā€, not local processing, so I was curios what that meant. Local vs remote processing is clear and I assumed not what the poster was referring to.

In this day of smart phones and the processing power even a raspberry pi gives you, there seems little reason for requiring remote processing other than audio voice recognition or image facial recognition. The simple analysis of a motion sensor turning on a light bulb or switch or activating a valve can easily be local. Lights coming on at dusk or motion detected in a camera. An ir remote control sensing a signal Turing down a dimmer. Even a phone app should be capable of acting as a switch/sensor on the LAN or Bluetooth.

If the v2 hardware processor isnā€™t up to that then Iā€™d say itā€™s an underpowered design.

They were referring to device types that are eligible to be processed locally.

A device type is code that formats the messages sent to a device. Only a few of the device types are eligible to be processed locally. Every device in your house will be assigned to one device type, but you have the option of changing which one you use.

It has been quite common for people to use a custom device type which has additional code that is not available in the official device type, perhaps to handle special parameters that are specific to that manufacturer.

One of the current issues for SmartThings customers is that if they do use a custom device type, the processing will no longer be eligible to run locally. So if you do want local processing, you pretty much have to go with the official device types.

In these forums, people commonly say their ā€œdevice is localā€ when they mean it is using a device type which is being processed locally.

Ah. Not being a ST user yet I wasnā€™t familiar with that terminology. Itā€™s disappointing that the user device types that communicate to local hardware that doesnā€™t have high compute requirements are not supported for local processing. I wonder what Samsungs reasoning for that choice is. If I have a water main shutoff valve I need to customize and I cannot get local processing to ensure a water detector under the dishwasher or wash machine will trigger the shutoff itā€™s a problem.

1 Like