Will a PXIe-8245 Ethernet Interface Module work in place of a typical managed Ethernet switch? Is so, advantages?

[rds2112]

This might belong on a hardware forum, but my need arises more from a LabVIEW standpoint: I have a Cisco gigabit managed Ethernet switch where I access 4 things. I have all four ports mapped in MAX, everything works fine, but there are two issues: 1, We'd like to keep the system inside the PXI chassis, and 2, most importantly for me, I'd like more LabVIEW access to the ports, especially in the form of aliasing, separate tasks, have LabVIEW test speed and connectivity before running the device hooked to a port, etc. Is the PXIe-8245 going to allow these features over and above the Cisco switch?

[Kyle97330]

I'm pretty sure that this device is just the equivalent of having 4 Ethernet ports on your system.  No switching, managed or not.  You'll just configure and use each one as if it was a network card.

[ZYOng · Accepted solution]

PXIe-8245 is just a peripheral module with four Intel® Ethernet Controller I210-IT in the PXIe form factor. It has no difference from other ethernet ports. All the communication with the hardware is done via OS. You cannot use any API from NI drivers to communicate with the hardware. That's the reason why NI MAX only displays the hardware but the configuration including the IP setting needs to be done from the OS side (Network and Sharing Center for Windows, ifconfig for Linux).

On the other hand, LabVIEW Internet Protocol API is just a wrapper on the OS socket API, with basic functionality. You might get more flexibility using socket API with C programming or third-party software.

[rds2112]

---

@ZYOng wrote:

....    You cannot use any API from NI drivers to communicate with the hardware. ... configuration including the IP setting needs to be done from the OS side..

---

Well shoot. Not *quite* what I had in mind. But, when using a network switch, you have to use the included (in this case Cisco) software for the setup . Perhaps, with the PXIe-8245, I can just rely on (MAX + OS) and not (MAX + some Cisco program + OS). Thus, removing one component from the setup procedure. It seems the driver that comes with the PXIe-8245 must have some value, else I'd just put a $99 Ethernet switch in an empty PXI module and be done with it.

[rolfk]

---

@rds2112 wrote:

---

---

Well shoot. Not *quite* what I had in mind. But, when using a network switch, you have to use the included (in this case Cisco) software for the setup . Perhaps, with the PXIe-8245, I can just rely on (MAX + OS) and not (MAX + some Cisco program + OS). Thus, removing one component from the setup procedure. It seems the driver that comes with the PXIe-8245 must have some value, else I'd just put a $99 Ethernet switch in an empty PXI module and be done with it.

---

PXI(e) is a little more than a standardized case around some random hardware. The board integrates as fully integrated PCI bus hardware interface into the system, similar to a PCI(e) in your PC. Except PXI is much more stringent about the requirements for such a module. Allowable temperature range, maximum power consumption per slot, air flow requirements for cooling, extra monitoring lines for the PXI backplane to measure things like temperature and power consumption. Designing and testing such a board is a lot more complicated than dropping a standard LAN switch chip on a board, arrange some connectors to be placed on the board and routed to that chip, a few decoupling capacitors and wrap some metal case around the whole, add and extra external power supply and replace the logos in the OEM software package that you can buy for that chip with your own.

In addition, Cisco and friends produce these boxes in numbers per model that a typical PXI board manufacturer doesn't even reach across a whole program of different PXI boards. This means that the entire more complicated design and testing cost needs to be distributed over a significantly smaller number of products.

Why would you buy PXI hardware then? It's usually considerably more robust, much more stringently designed and tested and its expected life time is normally way beyond 10 years, where as a lot of standard PC hardware are generally designed to last a little longer than the warranty time when used in an industrial environment and anything more you get out of them is luck.

You also get an all in one housing solution that can be easily put into a 19" rack or similar without a lot of interconnect cables that will sooner or later go bad for sure.  

[rds2112]

---

@rolfk wrote:

...

Why would you buy PXI hardware then? It's usually considerably more robust, much more stringently designed and tested and its expected life time is normally way beyond 10 years, where as a lot of standard PC hardware are generally designed to last a little longer than the warranty time when used in an industrial environment and anything more you get out of them is luck.

---

Hi rolfk, it's nice to see you again, and thanks for your excellent response. The snippet I have quoted from you above is well enough to justify the PXI version of a switch when channel count isn't too high.

Through research on this subject (Ethernet) which I am sorely lacking in knowledge, I have found that an Un-Managed switch will be fine for my application. I could have sworn there was a solution that provided aliasing of the IP addresses and LabVIEW control, but what I think I was looking at (this is going back 10+ years) was a mDNS-enabled solution where wrapper VIs were used, likely calling DLLs. I recall VIs that were able to reset a port, make it sleep, check it's connection (ack and nak), things like that. This obviously wasn't just a product of some PXI ethernet switch with magical drivers (as I had imagined), but rather was the result of a lot of work someone had done. This person (or persons) had written an API that fooled me. Good job I recon.