‎09-09-2013 09:49 PM
‎09-09-2013 09:55 PM - edited ‎09-09-2013 09:56 PM
Bad Idea.
First, what do they exactly mean by a virtual machine?
Virtual Machine often means running another operating system inside of the PC, like running a version of XP withing Win7. But in that situation, you still have a PC.
Another type of Virtual Machine is something like Citrix. Almost like going back to the old days before PC's where the person had a dumb terminal that communicated with a mainfram computer. Instead of deploying a real PC with a real hard drive and memory, it is a trimmed down box with display, mouse, keyboard, and a network connection. The OS and programs are running on a server.
If this is what they mean, I have no idea how that could work for you. Usually in a lab situation, you are dealing with real hardware, real signals and using real data acquisition cards. There is no way you can "virtualize" all of that.
Tell IT to leave their hands off your PC's and tell them to go back to their job of resetting passwords for people.
‎09-09-2013 10:01 PM
They want me to use a dumb terminal,with the application running on a server... and they want to access all of the instruments through the network.
‎09-09-2013 10:16 PM
Will you be able to communicate with all of the instruments through the netork?
Do they have a good reason for doing what they say they want to do?
They are going to put themselves out of a job. Once you have all your stuff virtualized on a network, the next step is them losing their jobs as they outsource network administration to India.
I think these virtual machines are just going to cause you more problems down the road even if you can get them working to some degree. I really think you need to fight these attempts by them. If they want to make this change, then it should be their resonsibility to prove to you that all will work properly before they try to implement their scheme rather than them forcing a change and making you try to prove to them that it won't work.
Here is one scenario I would pose that that works against their idea.
Do you have long running tests? Something where you set up equipment and the PC, and have it execute for a period of time continually collecting data? If something goes wrong during this test, is the test a waste, the test piece trashed? If you have a PC/real connection to your instrument, then even if the network goes down, the test should still be able to run successfully and the data collected. But now if you virtualize the PC and the network goes down, it basically takes down the heard of your program and your test is corrupted. Now they'll say they have back plans, redundant routers, blah blah blah. And it is usually rare that internal LAN's go down. But where is this server? Is it local to you or some other remote location? It is much more likely for the network to go down if you are jumping between buildings, states, even countries. And IT is notorious for needing to reboot a server, much needed security upgrades, blah blah blah, that will cause them to take down the virtual machine server. They may even promise that they'll notify you before they do such a step so that you can finish your test and make sure everything is clean and shutdown. But guess what, they never do.
‎09-10-2013 02:13 AM
@RogerMJR wrote:
They want me to use a dumb terminal,with the application running on a server... and they want to access all of the instruments through the network.
Can the instruments be accesses through network? If you're having any sort of pci-card or similar it probably wont work. If it's USB-connected it could work but will probably cause issues. If stuff is network connected it will probably work, assuming timing isn't of importance and so on.
All in all, lab equipment != office equipment and shouldn't be handled the same way. Have them test and prove it works.
/Y
‎09-10-2013 05:40 AM
My company's IT services are provided by an outside company that puts heavy restrictions on what we can put on "their" network, but they have no control over what we put into production. As much as it hurts to not have it on the network, it feels so nice not to have to beg for administrative access to "my" computer. Perhaps having a stand-alone computer is an option for you?
‎09-10-2013 09:51 AM - edited ‎09-10-2013 09:51 AM
@Yamaeda wrote:
@RogerMJR wrote:
They want me to use a dumb terminal,with the application running on a server... and they want to access all of the instruments through the network.
Can the instruments be accesses through network? If you're having any sort of pci-card or similar it probably wont work. If it's USB-connected it could work but will probably cause issues. If stuff is network connected it will probably work, assuming timing isn't of importance and so on.
All in all, lab equipment != office equipment and shouldn't be handled the same way. Have them test and prove it works.
/Y
Of course you can explain to them how awesome they will look when you aren't able to access the tester for an emergency shutdown and you have to call the fire department to put out the blaze.
‎09-10-2013 10:00 AM
You need to be careful on your virtualization software. If you use kvm, it provides the hooks that allow you to use the PCI busses. It provides information on how to write drivers so that the virtual machine can run the hardware directly. USB is an issue with regard to the Windows type operating system. You should in general consider using Windows vista or later to run on your virtual machine. XP has some bugs in its drivers making some USB equipment problametical with respect to machines and hubs.
Another consideration you need to determine is the type of response you expect on the virtual machine. Response will be considerably slower in a virtual machine, so the equipment you are looking for needs to have buffers. As a development system it probably does not matter.
The virtual machine running on another machine has to have careful resource issues. One of the things you need to know is how they do your scheduling etc.
‎09-10-2013 10:04 AM
IT departments are frequently completely ignorant in these regards. You will probably be forced to go through the experience of failure. I guess there isn't much you can consider in your LabView efforts other than describe to them which instruments need to be connected by which interfaces and then let them find out how to achieve it and maintain uninterrupted and safe operation.
Cheers
Edgar
‎09-10-2013 11:03 AM
@RogerMJR wrote:
They want me to use a dumb terminal,with the application running on a server... and they want to access all of the instruments through the network.
I would use the opportunity to design a test station using cDAQ/cRIO and LXI.
If management doesn't like the prices, you design a more traditional system and IT has to accept it.
If management does accept the additional costs, you are the winner!