Follow Us.
Become a Fan! 
There is no doubt that an entire thin client system relies on the operation of a single machine - the server. All software for every terminal located throughout the facility has been moved to this machine, and, as each terminal logs into it, the integrity of the server becomes more and more important. The server is therefore often identified as the system's Single Point of Failure. If the server is not running, none of the attached thin clients will run either. Is this hubris really unique to thin client systems? Lets start by looking at the methods currently used by systems that use individual PCs on each line in the factory.
Consider a plant with 12 identical manufacturing lines. Each line produces product, and has various pieces of equipment that can communicate with a computer. Each line usually also has a computer located locally, which is used to display process information and control the manufacturing operation for that line.
The more complex equipment will usually have some type of high-speed communications interface. Either Fieldbus (Profibus, DeviceNet, etc.) or standard Ethernet. This communications line will feed into a dedicated PC that acts as the Data Server for the entire plant. Each device has a logical address, and the Data Server 'hits' each on of them in turn and then 'serves' that data for any other applications on the network. While it is possible that there will be the required I/O cards and drivers installed on each PC (enabling each line's PC to communicate with all of the I/O devices on that line), this is rarely done. The complexity and cost of installing all of the PC side I/O is prohibitive.
The simpler pieces of equipment will usually communicate via serial data. If this data is ASCII, then it will sometimes be read locally by the PC on the line, although many times that data will be converted (electrically) into either 422 or 485 signals or, in certain environments, be put onto fiber optic lines. These signals then travel back through the plant's wiring until they reach the same I/O server, where they are processed and made available to all of the line PCs.
Even if a single machine does not collect the I/O, there will almost assuredly be a single computer acting as a repository for the data collected. It is used to run MES (Manufacturing Execution Systems), or simply to allow operators to view past performance of the line.
In either of these two cases (single I/O Server or singe Data Server), one point of failure exists. Lose either the I/O or the Data Server, and the system will probably not be able to function at all, and certainly not in a manner which will allow reasonable operation of the line to continue. The interesting thing about this inherent weak link is that it is often not recognized as such. The Plant Manager sees that he has a PC and a PLC at each line, and when he sees that the PC is displaying the I/O information he doesn't realize that the actual data is all routing through a single server. Or when he sees the report of the production from the previous shift, that all non-current data is coming from a single data server.
So, we have shown how most installed systems that use distributed PCs have a single point of failure themselves - this still doesn't answer how to deal with the obvious point that when the server in a thin client system goes down, all of the clients become paperweights. How is it possible to minimize this risk?
The answer, it turns out, is actually another of the benefits to thin client computing.
In an ideal world, the computer at each line would never go down. To do this, each line would have to have a redundant PC. This is never done, primarily because of the tremendous cost that would be incurred to achieve this level of up time. However, making the server in a thin client system redundant protects the entire system. Now, for the cost of a single redundant server, all of the terminals throughout the plant inherit the server's reliability. Bring all of the high speed I/O back to this server, and the Data Server is redundant as well. Serial data with lower collection rates can be sent back to the server, or can be collected by the local thin client
For installations where each line needs to function as a completely isolated module, collecting all of the I/O and saving all of the data, the thin client system will probably not be the best fit. But in the majority of systems, a thin client system with a redundant server will give all of the benefits of the thin client architecture, with the added feature of increasing the total up time of the factory wide system.
For more information on ACP Industrial Thin Client computers, please visit our web site at http://www.thinmanager.com
To sign up for the E-mail newsletter go here: ACP newsletter signup
For an archive of past newsletter articles go to: ACP Newsletter Archive
Copyright © Automation Control Products, Inc. All rights reserved.
