Thursday, January 14, 2016

On the road again and IO bound no more

Okay, weird title but that is what happens when I try to get cute. Recently I was looking for an SSD to use with a Raspberry PI for my TCF Home Automation presentation in March . It would need to work with USB 2.0, not a huge amount of storage, just local sqlite and logs files basically, and it need not be very fast at all. I found what I needed, that was both inexpensive and met my requirements. In my search I kept running across very expensive SSD PCIe bus adapters. Because of the cost I ignored them. But then I came across this ACM article called: Non-volatile Storage - Implications of the Datacenter's Shifting Center While data centers don't seem very important when talking about HA it actually will be in some ways. What goes for the larger systems tends to eventually get into the smaller systems. And with the IoT that's a lot of smaller things. I'm guessing that it could mean that we'll see more cores and better IO in our HA/IoT devices. I worry about many things related to this kind of architecture. Things like service longevity, reliablity, security, privacy and the need to for being 'always connected'. I really don't see Comcast, Verizon or any of the other providers making that affordable. But we're still moving towards everything being in the cloud. My preference is towards more local control and I'm betting the compromise will be a hybrid. Well that's my predictions for the New Year.

Sunday, January 10, 2016

How not to build a home automation empire

My recent Camera blog entry got me to thinking (yes I know, always dangerous). What is really needed to get the HA/smart home/IoT really going with the consumer? Well I can quickly tell you what won't work (always easy to do), bad software! The immediate problem I had with the camera was that the only thing that could use the camera was my smart phone. I couldn't install it on my Android table, weird! And the app was painful to use (big picture, little screen, I'm just not sure how well this plan was thought through). Yes we are going in that direction where the cell phone is central to everything but I found I couldn't use it with anything other than the meshare app. I couldn't get the data (images/video) into any other shared service and it seemed pretty clear that it wasn't going to work with anyone else's products. I would be trapped into vendor lock-in. While I'm no fan of the cloud it is the direction we're going in, so working with existing cloud services is important. I don't need a new service for each appliance in my house. Now what was the password to the refrigerator? And you need to work with other devices. I don't need to be accessing a dozen different apps so I can check the weather, irrigation and the HVAC. HA is a suite of products and services. So if you are building the next big thing in the IoT put a lot of thought into integrating other popular cloud services such as IFTTT and perhaps think about providing an API so that that makers of the world can add things you haven't thought of. By making your products and service interoperable with other products and services (that also interoperate with other services and products) you're more likely to last longer than the flash-in-the-pan devices we're seeing. Oh and also make your products and services work. Insecure, buggy devices and services won't cut it either.

At the moment there are plenty of 'knick-knack devices' and very few 'wow that's useful devices'. With the integration of cloud services that could change but not until the industry changes its attitude. The short life span and up-selling by discontinuing a product or service has to cease. The reason no one likes Microsoft's phones is that they have a bad habit of upgrade by dropping support. Remember WinCE? Those that purchased those devices wish they hadn't. A lesson they quickly learned not to repeat.

The central premise of the maker movement is being able to take one thing add something else (usually something with software) and making something else. Give into that movement and I expect you'll end up with a better product.