Issue link: https://iconnect007.uberflip.com/i/817722
May 2017 • SMT Magazine 83 It might be better if we just try to promote the OML as a given. OML is already on the table and available to everyone. You can adopt it now rather than wait three years for all these guys to agree on something—who knows if the end compromise is worthwhile? We're engaged, we're working with some leading companies to enhance where we have limitations or where we need some additional information. We're getting good traction and good feedback from a lot of industry players. Las Marias: The Industry 4.0 and IoT are among the major discussions last year. Do you think every- one really knows what that is about? Manor: Probably not. At the end of the day, a lot of people want to buy an Industry 4.0 appli- cation or smart factory solution. It's challenging for companies to understand what to really do from a practical perspective and how to make it efficient. On the other hand, smart manufac- turing does have some concepts and talks about Lean solutions; it talks about having very good data, and making real-time decisions. The Industry 4.0 discussion provides two topics which are good to talk about: data and connectivity. How do I get data? How do I really read and collect real-time information? And the whole idea about big data analytics…once you collect the data, you have a significant amount of data—even if you're a small, two-, three-line EMS. If you collect all of the traceability and quality data from all of your machines, you need a big data solution. A traditional SQL da- tabase is not going to be able to hold that data, so you immediately fall into this big data seg- ment and you need a solution for big data. You will now understand that you need to solve your data collection problem. You need real-time data, but you can't trust the data fed by operators or collected manually. If you re- ally want to know what's happening on the shop floor, you need to collect the data from the machine and you need a solution for the data captured if you want to do some analyt- ics. People now are looking for these two ap- plications because they understand that this is going to get them into this smart manufactur- ing domain. Las Marias: Exactly. And the solutions are avail- able now, right? Manor: The solutions are available, but I think that if you go to most manufacturing shop floors around the world, this is not imple- mented. While it's available, the implementa- tion is challenging and I would say that prob- ably 80% of the shop floors around the world don't really have this kind of data collection. Most people still use traditional SQL databas- es; we see the volumes of data and that need to be constantly archived. You have to constantly flesh out data. If you want to run quick analysis reports, you can only do it for one or two months of manufacturing. It would be much better if you could run this analysis over two years of manu- facturing—but the data chunks are just too big. So you need to implement a big data solution. And I think that the majority, or 90% of com- panies, don't really have a big data solution for their PCB assembly. That's something that com- panies are starting to understand and are start- ing to shop around for. Las Marias: Why do you think the implementation is very challenging for them right now? Manor: Because of the proprietary connections. At the end of the day, to get the data from all of these machines, you need a lot of drivers, you need to support all of these proprietary interfac- es, and that's a challenge. If you have a mixed vendor environment—with different vendors for AOI and SPI and reflow and printers and pick and place and ICT—you can count 20 different drivers. Now, if you have even a mid-sized IT MENTOR GRAPHICS: CONNECTING THE MANUFACTURING ENVIRONMENT " If you want to run quick analysis reports, you can only do it for one or two months of manufacturing. "