Hello All.

This topic might be almost a conclusive topic for IoT2040. I appriciate all contributors if they have any work arounds.

Basically the IoT2040 has got a customised hardware version of Intel Galileo Gen 2 Board. As on date you could only have the packages from this link & all the manufacturing done in China (You can confirm it by the print on IoT2040 or even by hardcoded MAC address). Never ever try the forked versions as it has got certain license limitations with the long discontinued Intel Galileo Gen 2. I wonder how Siemens could still sell or keep this product active as the open source board has been discontinued long ago.

With all the technologies said & done a Simple ESP8266 or Node MCU (Arduino Original Made in Italy) would work much more flexible & happening compared to IoT2040. I wold get an industrial grade of those alternatives @ hardly $2 or €1.8 euros + duties & local taxes. Or even NVidia Jetson Series Industrial Grade Packages are far far matured technically in comparision. You can do the math how many you can buy in place of one IoT2040. Nevertheless we all know by now that IoT2040 has reached its maturity without justifying the term Internet Of Things. There are IoT2040PN based solutions also which are nothing more than unnecessary complexity. Nothing wrong in using open protocols on a third party device like Modbus over TCP etc rather than pressurising on Non confirming Profinet connectivity over Internet.

For Example We have a dedicated server hosted by Google where we have Ubuntu20.04 Lts Running with all open source packages like Apache 2, PHP, MySQL, Python3 etc running fine with plant data. We have Arduino Mega Based boards & Jetson Nano based boards both industrial grade & working with understandable straight forward IDEs like Arduino IDE, PICharm etc.On the other hand there are no community or professional IDEs to suffice this requirment for IoT2040 as an internet gateway. By the way, officially as on date except for Node-red & Ubidots there are no transparent methods to get connected to internet from a IoT2040 device for a user defined program on any platform. We have all the options to develop the required shell scripts or tweek the OS (Yocto-Linux) to enable this all. Apparently/comparitively it is simple to tweek a router to run our IoT scripts rather than finding a work around in IoT2040 OS. Hope the same is not with IoT2050.

The over exaggerated/decorated solution of Mindsphere is still not a good option when the Industries already have dedicated servers, database engines in place for Industry4.0 requirements. After a simple math, for one of our customers the Server side requirements with data backedup costed us as follows

GCP: (VM With e2-medium) - $6000 + taxes per annum for a 30GB persistant disk. (With Static IP for server no DNS, no mail server) 

After a complimentary 90 days $300 free period we could complete the development activities & put it on regular service with N1, F1 Micro version package which will cost not more than $150 per annum (One hundred & Fifty Dollars Only per year)+ $60 per year for a persistant SSL (Confusing though). + $150 per annum for Mail Server & DNS for practically infinite number of users each with a 15GB of personal drive space. Vola Industry4.0 achieved for a trillion dollor company for hardly $300 per annum.

AWS: Tried hard but compared with GCP by official sales representitive around $12000 per annum (Fully defeatured - 50GB persistant disk, no mail server, no DNS, no static IP for server). But finally dropped it as in market after the 2015 security breach with AWS, most of the F&B and Pharma companies are moving out of AWS forever & in India it is almost done.

Azure: (The main engine actually behind Mindsphere): $16000 If worked out directly for 20GB persistant disk & a Windows Server with a professional Server license, no DNS, no Mail Server.

Almost all the customers we have Interacted since 2011 would require the same. They have a stipulated business analytics to be done & they have all the documentation for Data Sciences to be implemented & dedicated server/cloud hosted in their infrastructure or bought out. None of the providers as on date would even offer a standard package anywhere close to the front end & backend requirement. If at all it has to be customised then "better get the engines & do the development as required inhouse" is the option that everyone will follow. Practically perfect for those who have an idea on technology. For those who dont know will keep beating around the bush with standard package offerings.

Coming Back to IoT2040, Connectivity with S7 PLCs is a third party tool (atleast the best tool i must mention). Official Examples are only limited with Node-red or Ubidots which I still dont understand what are the professional standards with such solutions. As a gateway it is simple & Strainght forward that Siemens do not offer any example or method to connect this device to user required servers over internet in any IDE(Platform). Concept of IoT is officially dead by now. With this background there could not be anything better with IoT2050 TI SOC AM6528 GP Dual Core or TI SOC Quad Core Version luckily still active. But once said TI(Texas Instruments) we are well versed with the complexities behind this hadware, Firmware & Softwares. Not an option at all that too when the hardware is customised, that too by Siemens.

In short The Siemens Hardware offerings for Industry4.0 with respect to IoT specifically mentioning IoT2040/IoT2050 are not a professional solution. As an academic project if supplied with complementary licenses & real world costs it is worth to just give a try.

What Say @bergmanu, @IoT2000_Moderator?