- Home » Features » White Papers
Which requirements do heterogeneous networks have for the switch
The switch as intelligent “stop light” in the data traffic (Originator: Indu-Sol)
How industrial communication succeeds reliably
Which inner city can afford to plan four traffic lanes next to each other: One for normal automotive traffic, one for lorries and buses, one for ambulances and one for the fire brigade? What this question has to do with industrial networks of today is the topic which we will discuss with the managers of Indu-Sol, Karl-Heinz Richter and René Heidl in our interview. We will explain to you how networks develop further, why the criteria change that the mechanical engineers base the selection of a network switch on and whether we need TSN and more bandwidth.
Editor: Herr Richter, your core business exists for 18 years in making a stable and reliable communication infrastructure possible for the production level. How has this task changed in the last years and where do we stand today?
Richter (Fig. 1): In the early days we had a lot of explaining to do. The Profibus was celebrated everywhere and no one wanted to believe that it also could have weak spots. We explained a lot: That there is something such as physical wear which can lead to problems in communication sooner or later. The practical experiences have confirmed our statements.
Fig. 1: Karl-Heinz Richter, Managing Director for Marketing & Sales at Indu-Sol GmbH. (Originator: Indu-Sol)
By switching to Profinet the communications landscape was not only changed, but also the problems that can accompany it. In the Profinet we are struggling primarily with sporadic events that are not reproducible, which can be simply acknowledged first of all and which actually seem to be unproblematic.
Editor: But these sporadic events are problematic anyway?
Richter: Yes, because data is also lost then. This becomes visible in a consistent network monitoring. And even more so: These sporadic malfunctions are always precursors for a looming network problem.
Editor: What else has changed with Profinet in regards to the communication networks?
Heidl (Fig. 2): As an argument for switching over to the TCP/IP-based communication it was promoted to us that the most varied applications can be integrated. While networks were homogeneous with Profibus, we suddenly were dealing with heterogeneous networks with Profinet. In the meantime many are back-pedalling and are routing several homogeneous networks parallel. (See box text "Heterogeneous and homogeneous networks") This development is, however, not really known to the most.
Fig. 2: René Heidl, Director of Technology & Development at Indu-Sol GmbH (Originator: Indu-Sol)
Editor: Why are so many going back to homogeneous networks?
Heidl: In brief: Because homogeneous networks are more stable, reliable and easier to monitor. But for that you have to monitor each network individually and to create interfaces to each of these networks.
Editor: And that significantly more laborious?
Heidl: Of course. Imagine that four traffic lanes are being constructed in your inner city: one for automotive traffic, one for buses and lorries, one for ambulances and one for the fire brigade. Who is going to pay for that; I mean for the construction as well as the maintenance?
Editor: No one. But what does that have to do with the communication networks now? Can't all protocols simply run on one "lane"?
Richter: Sure, but the stop lights are the core problem. Whereas with a stop-light system six cars can easily zip past the light, one lorry just starts up and then it is red again. Sooner or later that will lead to traffic congestion. Transferred into the network world the stop light is the switch that manages the communication protocols. If short (e.g. Profinet) protocols and long ones (e.g. visualization data) run through the same switch, then no congestion occurs, but rather data is lost in case the buffer is not large enough. But due to just that, the networks become instable and that leads sooner or later to communication failure and to system standstill in the worst case.
Editor: So the switch is the actual problem?
Heidl: Precisely. Coming from the homogeneous Profibus network, the selection criteria for the appropriate switch were often the housing size, the price, the protection class and as most important argument, the manufacturer. Whoever was employing a PLC from Siemens, Rockwell, Phoenix etc., always reached for a switch from this manufacturer as a rule. And that also made sense absolutely – in homogeneous networks. Because PLC and switch are perfectly harmonized with each other.
Editor: But in heterogeneous networks machine and system engineers should apply other criteria? Why?
Richter: Now we are dealing with a number of applications and communication protocols. Some are short and are sent frequently, others are long and come less often for that. A classic Profinet switch is optimized for example for short Profinet protocols. But if you run a visualization application now over the same switch, it is not optimized for that and problems arise. But we already talked about data loss etc. …
Editor: What is the alternative?
Richter: Well, either you just decide to route an own line for each application. That would get expensive during the installation as well as during the maintenance. Or you set other criteria when selecting the switch.
Editor: Which criteria are important when selecting a switch today?
Heidl: The bandwidth. From the start you should think about designing the backbone of a machine in gigabit, that is backplane capacity, data throughput, buffer size and buffer page size (for explanations see technical box 2) as main arguments for the selection of the switch instead of the name of the manufacturer. That is how it is handled for some time now in the IT field actually. Why the OT area (OT = Operational Technology) behaves completely different here is not comprehensible. Then one would have enough leeway though to reliably transfer the data of all applications in heterogeneous networks.
Editor: If I understand this correctly, then this means that you can either use a transmission rate of 100 mbps but then need TSN or use a transmission rate of 1 Gbps and then do without TSN. In both cases the actually available bandwidth would increase. In light of cost constraints one would choose gigabit then, because TSN components get significantly costlier due to the required time accuracy. Is that true?
Heidl: Those are your conclusions. I don't want to delve on that any further, but I can't contradict that completely.
Editor: If the world of industrial communication has changed so much, allow me to pick up at our initial question once more: Is the field of activity of your corporation different from 18 years ago?
Richter: Yes. Over the years we have evolved from a diagnostics provider more and more to an OT network supplier. However, one corporate branch will not completely replace the other. Nevertheless we have thoroughly expanded the hardware area. A new development in this area is our Managed Industrial Ethernet Switch PROmesh P10 (Fig. 3).
Editor: How does PROmesh P10 differ from other industrial switches?
Richter: It is compatible to the new standard "PROFINET 2.4 via TSN". And then there is much diagnostics data in the infrastructure of the switch which actually only needs to be made visible. That is why we offer in addition to the shielding current measurement, which was already implemented in the predecessor model, now also a performance diagnostics that is carried out so intelligently that the switch signals a warning against the total communications breakdown. Thus preventative maintenance of the network becomes possible.
Editor: Why are these diagnostics necessary?
Heidl: In case the quality value “CRC error” accumulates, this indicates performance problems and the question for the cause comes up. Either EMC couplings, defective lines or a “weakening” end device can induce these errors. If the P10 shows errors in connection with high leakage current values, then the cause for error lies in the area of the EMC. If the P10 shows errors in connection with poor cable quality values, then either the connection is defective or the connected end device.
Editor: How do you realize the cable diagnostics?
Heidl: During the cable diagnostics, the form of the real signals are evaluated and further processed by a data matrix. Based on the transmission quality of the signals clear clues to the cable quality can therefore be made. Since many experiments and also AI-like evaluations have preceded this solution, we see an absolute novelty on the market here.
Editor: Final question: What is the aforementioned performance data of your switches?
Heidl: Good question. We currently offer a backplane capacity of 51.2 Gbps, a throughput of 19.3 Mpps. The buffer size is 2 Mbit, with a buffer page size of 256 bytes.
Editor: Thank you for the interview
Technical box 1: Heterogeneous and homogeneous networks
Homogeneous networks are understood in the OT area as uniform networks in which the manufacturer of the controller also provides the network infrastructure and only one single application (i.e. Profinet for example) is used in the whole network. In heterogeneous networks on the other hand various applications run in the same network. Furthermore, the controller and the network infrastructure can be procured from different manufacturers.
Technical box 2: Explanations of terms
Backplane capacity
The backplane capacity is a measurement for what data amounts a switch can process per time unit. It is specified in bps.
Data throughput
The data throughput describes how many telegrams a switch can process per time unit. Its unit of measure is pps (packages per second).
Buffer size
From the buffer size you can draw conclusions on the data amounts that a switch can store temporarily (buffer). It is specified in bit.
Buffer page size
The total memory is divided in individual memory pages. Per page, a maximum of one telegram can be saved. The buffer page size specifies the fixed memory size of the memory page and is specified in byte.
About Indu-Sol
The reliable and trouble-free communication is the guarantee for a consistent production. Indu-Sol offers a comprehensive product portfolio and tailored solutions for that to improve the reliability and stability of your networks. The corporation thus sees itself as a comprehensive partner for industrial networks. Beginning with a network consulting/planning and the delivery of components (infrastructure and diagnostics) and all the way to the technical on-site service - this means support for the topics commissioning, maintenance, service and training.
For more information, please visit http://www.indu-sol.com/promesh-p10-en
