I’ve also tried using the SmartThings HA Integration, but that works horribly. And also, not what I really want.

Light is a Lithonia CPX 2x2. I’m fairly certain it can support ZigBee, although there is a tiny chance it’s really using BTLE.

Anyone have any experience with these or know of a not supper complicated way to determine which protocol this is really using?

I'd love to see a world where hosting all of your apps and data in your own closet would be simple and common.

I've also been noticing that newer deployment models are trying to leverage a variety of different providers for different purposes - both to save on cost but also as a redundancy measure.

Still, I think the trend is early, but I think it's certainly becoming a trend.

While I don't think the "fediverse" has gained too much more momentum yet, I'm certainly seeing more discussion around "personal data ownership" than I had in the past.
And platforms such as anytype.io are also gaining some traction.

I'm seeing an uptick in non-traditional cloud deployments, more adoption of federated networks, and also data platforms where you own the data.

Basically the short version of they story always ends up being: pay attention to the details and the interaction patterns at a more global scope, and not just in a single service. The single unit, service, query usually isn't what causes the problem. It's all of the interactions at scale.

There is also the other end of the spectrum - services making calls to other services. Where a single service call is generally quick. But then you chain together 3, 5, or 10, and things start to slow way down.

Or a bulk query typically gets called with maybe 50 IDs together, and it's alway ok, until that creeps up and ends up being 1000, or 2000, and it's not longer ok.

A simple database query isn't problematic, but a simple database query called 1,000,000 time is. Except that can get hidden in all of the layers above the data layer, and no one notices or realizes

Here's what I spend most of my time really educating people about, and what ends up being most import to wrap your head around: There usually isn't a single place in the code that will kill performance (there are, but it's rare). It's more about the interaction of everything together.

When everything goes bad, make sure you can trace the steps that led to the outage or the breach or the failure. Without very good logging and monitoring it can be impossible to tell what went wrong and what caused it.

An organizations plan usually isn’t tested and usually isn’t walked through in any detail. Until you actually try to determine what is going wrong or who got into your system, the plan will fail.

I’ve seen a few different alternatives from various companies:

- Build a small app that’s business related (pre-defined, same for each candidate)
- Give a presentation about something in tech you’ve done
- Pair code a small feature or or fix

What are alternatives that you recommend?

Or maybe it's always cheaper and more efficient to just buy from third parties?

Any thoughts?

And then we got SaaS and "reusable services"...

Again, third parties making lots of services they will sell you, although this time there are some good ones...

But, will we ever get to a point, where a company will make something reusable that they can then build countless products on?

But a bunch of software vendors made a bunch of reusable components that they sold a bunch of (and they usually weren't that great)