How do you deal with latency issues arising from event driven architectures?

Top Answer : It's a great question. We're living this. I mean, this falls right into the germane of sensor fusion and the applications and use cases that are arising in all kinds of different verticals, from intelligent traffic systems to next generation manufacturing to the automated worker safety, etc. What we found interesting in working with different customers in these kinds of verticals is... regardless of whether it's video as a sensor, or audio as a sensor, or radar information or asset record information, it's just data. At the end of the day, by making that data available on a data net, what we can do is open it up to really intelligent people who can write event driven code that can make intelligence out of this. I can't make intelligence out of it. I can't tell the difference between an audio and a video data file on our network, but intelligent folks who can write algorithms and make decisions based on what it's interpreting or what it's reading in real time... they can do that. Creating these data streams from the machines and devices to the functions that are intelligent code doing something tangible and real and sophisticated with the data….that is what this is all about. So that, to me at least, is where the concept of distributed in the context of Edge is where the rubber hits the road, so to speak.

12 views
2 comments
2 upvotes
Related Tags
Orange Monitor
Software
It's a great question. We're living this. I mean, this falls right into the germane of sensor fusion and the applications and use cases that are arising in all kinds of different verticals, from intelligent traffic systems to next generation manufacturing to the automated worker safety, etc. What we found interesting in working with different customers in these kinds of verticals is... regardless of whether it's video as a sensor, or audio as a sensor, or radar information or asset record information, it's just data. At the end of the day, by making that data available on a data net, what we can do is open it up to really intelligent people who can write event driven code that can make intelligence out of this. I can't make intelligence out of it. I can't tell the difference between an audio and a video data file on our network, but intelligent folks who can write algorithms and make decisions based on what it's interpreting or what it's reading in real time... they can do that. Creating these data streams from the machines and devices to the functions that are intelligent code doing something tangible and real and sophisticated with the data….that is what this is all about. So that, to me at least, is where the concept of distributed in the context of Edge is where the rubber hits the road, so to speak.
0 upvotes
Orange USB Stick
Software
To your point about latency, that's a huge part of this. There's a couple ways to look at that. One is, not every event has a latency issue. In fact, my argument would be when everybody's doing event driven, when it's really standard to just send data instead of having an API, a huge percentage of the events are going to be much slower with much less latency issues assigned to them. That being said, I think you already see when you look at AWS and what's happening with their serverless portfolio and what they're beginning to add from an operational perspective that there's a need for that layer of platform capability to behave like an operating system or like a compiler. It has to optimize, and you will see the ability to make really intelligent decisions about where to place things and then where to route requests from the location that the request is coming from. It’s incredibly important to begin to build that layer of intelligence that makes smart decisions.  When routing a subscription request, for example, where do you place the data, where do you connect the subscription to get access to the stream, what do you do to optimize the network between the end user and the stream, between the stream and between the core system of record data sources, whatever that may be. That to me is a big, big part of the opportunity that comes forward. Cloud providers, frankly, have an advantage there. But if it's got to be done at the Edge, they're not really doing it at the Edge yet, and so there's an opportunity from an Edge computing perspective to push that out from a central data center model to ... yeah, we're within a hundred miles of your user's call no matter where you might be, in an ideal kind of fantasy world.
1 upvotes