Scattered Cloud: Amazon makes a timely play for IoT edge computing
via Flickr © peter_stanton86 (CC BY-SA 2.0)
- AWS makes its 'Greengrass' edge computing software 'generally available'
- Amazon gets the jump on Microsoft Azure at the edge
There’s a an important component in human physiology called the Autonomic System which more or less does for us humans what Edge Computing is designed to do for the cloud. The relevant part of the Autonomic ‘system’ is the ‘sympathetic nervous system’. This is the thing a doctor is tapping into when he or she hits you on the knee with that little hammer - it’s designed to trigger your "quick response mobilizing system". This allows you to jump into action without actually engaging the brain to think about it first.
The industry is now more convinced than ever that the same sort of function is going to be crucial for the IoT and M2M. All are agreed that The Cloud is going to be a key enabler of IoT, but components working quickly at the edge of the network and on the customer premises (in many cases) are going to be necessary as well and those who want to be key players are moving quickly to fill the gaps.
This week Amazon Web Services made a concerted move into IoT edge services with the launch of AWS Greengrass which is AWS’s solution to extend its cloud computing services out to the edge of the network. In doing so it’s chalked up a win against Microsoft in particular which has also been active with its Azure IoT effort. It has announced an edge component but it’s not yet available, which the experts seem to agree, has given AWS and Greengrass a head start.
Greengrass was announced last year, now it’s going to be generally available and supporters like Lenovo and Qualcomm are synchronised supporters.
Greengrass gives AWS users the tools to process data on local (to them) hardware before (or instead of) forwarding it to the AWS cloud for storage and further processing and analysis.
So Greengrass is a way to dynamically split processing workloads between edge devices and the cloud to meet application requirements. This could involve an expectation that direct comms links between the edge device (which will likely be collecting data from multiple sensors) might be intermittent (think ships or remote sites).
Or, it could be that the application involves industrial control tasks which means that it has to respond instantly to incoming data (no time for sending data off across the network to a remote server in the cloud). Or the edge device may be collecting and processing vast amounts of data from its own attached devices, only some of which needs to be forwarded into the cloud for further analysis. In this last case the edge device will act as a filter.
And of course the application could involve a combination of all these requirements, all performed with the requisite security.
So at its most useful, Edge Computing will be tasked to provide a machine equivalent of our own human ‘quick response mobilizing system’ when useful and required required. The Edge Servers can therefore be put in charge of multiple locally situated sensors or ‘things’, and can control them to filter out unnecessary data before it starts journeying to the cloud.