(republished on linkedin)

Over the past years most companies developed a cloud-first strategy and migrated IT systems from an on-premise setting to the cloud. Still, many of these projects are ongoing or are newly started. There are various reasons to drive for the cloud - a few examples are:

  • A company does need to invest into hardware to run their own applications. The capital is not tied up.
  • Separation of concerns: A company needs to have their applications up and running but is not really interested in the underlying hardware. The applications provide business value and probably differentiate the company from competitors while the hardware is a resource.
  • Making use of the expert knowledge of cloud providers instead of reinventing the wheel.
  • Time to market is faster.
  • Flexibility as resources can be simply booked when they are required.

Generally, the cloud is a highly optimized and integrated combination of computers and services. However, in some settings the cloud has some disadvantages. External applications communicate with the cloud via Internet technology. This means that system landscapes and applications are subjected to the limitations of the communication layer. These are:

  • Latency
  • Bandwith
  • Connectivity

In highly distributed environments, where a lot of communication takes place and data needs to be exchanged fast execution logic and analytics in the cloud might not be the best solution.

Edge, fog and mist

The scenario described above can be found quite often in IoT and Industry 4.0 scenarios; especially, if you talk about production and machines. Latency can be crucial to stop machines immediately if some technical mismatch is found. A complete connection loss would result in a downtime of production which is extremely expensive. This is why there is the tendency to handle data and to execute logic as close as possible to manufacturing. Edge, fog and mist are terms which describe the tendency to have computation power in manufacturing. They have more or less the same meaning: an additional layer between machines, sensors, devices and the cloud. This layer contains software to run, for example, stream analytics, data collection, communication and orchestration.

Industrial server at S1

For prototyping purposes, we recently bought an industrial server which is protected against dust and vibration. Therefore, we are now able to work on industrial scenarios which require more processing power or need to establish the edge concept.

industrial-server

The industrial server has 16 GB RAM and an Intel iCore5 7200U which is sufficient power to run different servers or IoT edge software. Additionally, it has many possibilities to connect to different kinds of networks.

Quickly tested

As an initial test we installed CentOS (Linux) as operating system and the log management software Graylog.

graylog

Graylog is a great tool to search for log entries from different and distributed applications. Moreover, it has a great visualization and alert functionality. Currently, a raspberry pi (single-board computer, IoT device) sends log messages with the current humidity and temperature to our industrial server. Graylog aggregates, stores and visualizes the data so that we have an overview of historic temperatures. Furthermore, we are able to define notifications and alerts if specific thresholds are reached.