Edge computing

Hi,
I need to deploy, in relatively near future, one project which main purpose is to test Edge computing capabilities, understand pros and cons, and get us a little bit more experience with mentioned theme.
Is there anybody here with real life experience, who could give me some directions upfront?
Any information would be appreciated.
Information I could find is mostly just advertising junk, that wrote IT guys without understanding of Operation Technologies and Factory Automation.

How do you define "Edge computing" in the Operation Technologies and Factory Automation contexts?

Hello, I guess it really depends on what you want to achieve. I think many of us are on a similar journey of discovery. Edge computing might mean different things to different people e.g. Remote data gathering vs decentralised processing vs Automation and control over the internet.

I have several e W O N devices deployed around the world that are sending back key metrics via MQTT. It is also possible to send data back to the machines via the same mechanism.

If your interest is in cloud based data gathering and dashboards then I can point you in the direction if I X O N https://www.****.cloud/ I have one of these units setup on an R&D machine.

Today I have requested a quote for a relatively new device from OPTO22 called Groov RIO https://www.opto22.com/products/groov-rio and this looks really interresting as an edge device.

If you want to see the sort of thing that can be achieved with distributed control ssytems then take a look at this video from Cirrus Link: https://vimeo.com/257174334

Nick

drbitboy said:

How do you define "Edge computing" in the Operation Technologies and Factory Automation contexts?

Click to expand...

We want to connect equipment on factory floor to Edge computer which will be connected to the Cloud. Then we want to test different solutions/options and see benefits and drawbacks.

This is what we use for all of our "Edge of Network" devices, connects directly to Ignition Scada for "Subscribe/Publish(MQTT)" in realtime.

https://www.maplesystems.com/product/modelname/cmt-g01

Manglemender said:

Hello, I guess it really depends on what you want to achieve. I think many of us are on a similar journey of discovery. Edge computing might mean different things to different people e.g. Remote data gathering vs decentralised processing vs Automation and control over the internet.
Nick

Click to expand...

I want to continue "journey" I've already initiated. As I have deep knowledge about Automation, Remote data gathering, Security I do want to take advantages Cloud platforms (AWS,GCP,Azure,IBM) and understand potential problems (and I already see them a lot) and find countermeasures.

damica1 said:

This is what we use for all of our "Edge of Network" devices, connects directly to Ignition Scada for "Subscribe/Publish(MQTT)" in realtime.

https://www.maplesystems.com/product/modelname/cmt-g01

Click to expand...

I've used that product, but under different name (Weintek). I'm interested in few steps forward. There are solutions that offer IPC with two operating systems, one is Win10 and other is RTOS installed parallelly, with many different ports (Ethernet, Serial, USB etc.), that are build to last and to be reliable. For example - you can restart Windows from RTOS, and they have parts of memory that are shared.

I put in a project using Ignition Edge that works pretty well. It was to gather data at a facility across the state and send it back to our ignition servers in our office. The system worked fairly well. There was some hiccups with getting the correct settings in order to get communication and data retrieval working but it worked just fine in the end. It was fairly cheap as well- few thousand dollars with a physical industrial pc and enclosure. Pros in my case were just in the ability to set it and forget it cons were that it was a pain to set up the first time. I am sure that I could go back in and set it up faster now that I have an example though.

jakeparsons03 said:

It was to gather data at a facility across the state and send it back to our ignition servers in our office.

Click to expand...

Were there some data analytics that was automated?

jakeparsons03 said:

cons were that it was a pain to set up the first time. I am sure that I could go back in and set it up faster now that I have an example though.

Click to expand...

What were the main problems?

https://inductiveautomation.com/ignition/edge

I'd take a look at Ignition, I've only used it at a server level but the edge stuff looked interesting. I no longer work with Ignition but I would 100% use it again.

Take a look at the Purdue model.

There are many software and hardware available. In short, Edge computing connect the plant via one of the cloud protocol such as MQTT to one of the cloud server. Once data is in the cloud then everyone who have access to that portion of the cloud can get the data.

You already got some good suggestions. I would also look at the Redlion Datastation.

harryting said:

Take a look at the Purdue model.

There are many software and hardware available. In short, Edge computing connect the plant via one of the cloud protocol such as MQTT to one of the cloud server. Once data is in the cloud then everyone who have access to that portion of the cloud can get the data.

You already got some good suggestions. I would also look at the Redlion Datastation.

Click to expand...

The only worry I have about cloud and everyone going to it.

What if it goes down? Edge buffer, site server buffer, company buffer server and then cloud I guess. I just don't understand going to companies that have an obligation to keep your stuff until they don't. Bankrupt cloud company?

VAN said:

What if it goes down?

Click to expand...

It comes down to 2 things in my estimation. What kind of data is it and relative risk.

Some type of data you do not want to only be on the cloud. You want it to exist locally. The type of data, say, a product reciepe, for example. You want the plant to run even if internet connection is lost.

As for relative risk of data in the cloud vs locally. My mind is slowly adopting to the cloud. There was a great article I read just last week about someone who lost his data on a hard drive and the author talk about how risky that is in retrospect. If one critically think about it. there are tons of risk with your own data-center as well. Beside the physical risk of fire and flood, you also have risk of ransomeware (yes, that exist for cloud data as well). What I'm saying is that cloud may not be any more riskier than local data storage and at the same time provide tons more functionality for others to access it.

harryting said:

As for relative risk of data in the cloud vs locally. My mind is slowly adopting to the cloud. There was a great article I read just last week about someone who lost his data on a hard drive and the author talk about how risky that is in retrospect. If one critically think about it. there are tons of risk with your own data-center as well. Beside the physical risk of fire and flood, you also have risk of ransomeware (yes, that exist for cloud data as well). What I'm saying is that cloud may not be any more riskier than local data storage and at the same time provide tons more functionality for others to access it.

Click to expand...

Fair enough, I was just thinking long term storage and having set timelines needed to retrieve archived data. So if a cloud went down for a week (probably unrealistic) and you only have a 2 day window to show the archived data.

Depending on the alphabet people (FDA, FBI, etc.) that might put you in hot water. I guess depends on industry.