Experimental Physics and
Industrial Control System

"Konrad, Martin" <[email protected]> · Wed, 15 Jul 2015 14:24:17 +0000

On 07/15/2015 07:56 AM, Benjamin Franksen wrote:
> On Tuesday 14 July 2015 12:16:38 Pete Jemian wrote:
>> Is your analysis of "expensive" based on initial cost, total life
>> cost, or perhaps some other criteria?
Our experience is that maintenance cost can be significantly lower for 
Linux machines. You can use all kinds of modern development, deployment 
and monitoring tools that are not available for VxWorks/RTEMS. You can 
leverage the advantages of a multitasking OS and powerful multi-core 
hardware (if needed).

> Also note that Linux is not a real-time system, which may or may not
> be a problem depending on the application.
Our strategy is to move as many real-time tasks as we can to PLCs, motor
controllers or FPGAs. In most cases this means we can run standard Linux 
on our PCs.

> It may also be a potential security risk, given that the amount of
> readily available tools for an attack is much larger on Linux (or any
> other general purpose OS like Windows or MacOSX), compared to VxWorks
> or RTEMS.
You are right: The amount of malware targeted on Linux is bigger than 
the amount of malware targeted on RTOS. However, the number of attacks 
on industrial devices has been grown dramatically within the last years. 
Also note that security issues affecting a Linux base installation are 
usually fixed much faster and can be deployed much easier/faster using 
tools like share libraries and software packages.

> And cutting something like Linux down to the essentials
> needed for running one or more soft IOCs (in order to avoid these
> risks) is not something I'd take on lightly.
On the other hand most VxWorks/RTEMS machines I've seen so far were 
using technology like telnet (instead of SSH), NFS (no encryption, 
single point of failure) and did not provide any useful IT monitoring 
(fan speed? temperature? free disk space? ECC errors?)... RTEMS is also 
lacking support for time synchronization using PTP which will be the 
default for FRIB.

Our answer is a combination of real time tasks in PLCs/motor 
controllers/FPGAs + high-level processes on Linux leveraging the latest 
tools to help the administrator maintain it.

-Martin

-- 
Martin Konrad
Control System Engineer
Facility for Rare Isotope Beams
Michigan State University
640 South Shaw Lane
East Lansing, MI 48824-1321, USA
Tel. 517-908-7253
Email: [email protected]

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System