Ethernet Subnet Mask

I was hoping to get some feedback on how a particular network is planned to be setup and see if you all think it is wise, or not.

This plant has approximately 20 PLC's (AB CLX,CpLX), but will grow over the next few years. To "tie" everything together the IT department wants to make the IP scheme 10.x.y.z with a subnet of 255.0.0.0. They don't want to use a layer 3 switch or assign gateways. The plan is to use unmanaged switches throughout the plant. I guess my main concern is won't this setup cause a lot of unwanted traffic since the subnet is letting everything through and there are no managed switches?

I'm not strong in networking, so please let me know if I have left any useful info out of my explanation.

Thanks,
James

Useful info would be if the PLC's are communicating with each other via Ethernet. Other useful info would be if each PLC is controlling various other devices via Ethernet. You've listed 20 Ethernet devices in your post, with no information regarding what they are communicating with. The OTHER devices, (remote i/o or drives communicating via Ethernet) as well as if/how the communication BETWEEN the PLC's is the critical piece of information missing from your post. Good job trying to stay ahead of the game!

As far as switch selection I would think about using managed switchs even if it is for the diagnostic functions.

Send this to your IT guys so they know what they are dealing with.

http://literature.rockwellautomation.com/idc/groups/literature/documents/td/enet-td001_-en-p.pdf

bulletin blues said:

Useful info would be if the PLC's are communicating with each other via Ethernet. Other useful info would be if each PLC is controlling various other devices via Ethernet. You've listed 20 Ethernet devices in your post, with no information regarding what they are communicating with. The OTHER devices, (remote i/o or drives communicating via Ethernet) as well as if/how the communication BETWEEN the PLC's is the critical piece of information missing from your post. Good job trying to stay ahead of the game!

Click to expand...

Some of the PLC's will communicate by produced and consumed, and a select few will do messaging. I forgot to mention there are a couple of ML1400 also. I don't know exactly what devices are on Ethernet, however, the two systems I am dealing with each have two CLX chassis's, one 13, the other 17. Both chassis's have 4 ENBT cards each. One ENBT is for HMI's and programming termials, the 2nd ENBT card is for IO (Point I/O), the 3rd ENBT card is for Encoders, the last ENBT is for drives (PF40 22-Comm-e, about 20 drives each system).

The reason why I ask this question is for when they start adding a bunch of "stuff" is there any risk of having bandwith issues? For sake of argument, lets say there are 20 CLX chassis's each with 2 ENBT cards, the first ENBT has 20 drives, the second has 5 Point IO chassis's with 10 discrete cards each. So, that is a total of 400 drives on the network (granted separate ENBT), and 100 AENT adapters (also on separete ENBT). Using this example, with the fact that everything is tied to unmanaged switches, is there potential for problems, or do the ENBT cards separate enough on their own to avoid issues?

Hopefully this all made sense. Let me know if it didn't

Hydro Power Guy said:

Send this to your IT guys so they know what they are dealing with.

http://literature.rockwellautomation.com/idc/groups/literature/documents/td/enet-td001_-en-p.pdf

Click to expand...

Thank you. I will forward to them.

James

IMHO each major process line should be its own network with a vlan between process lines that need to talk to each other.

This requires a layer 3 solution to be able to manage them all from the same engineering computers and maintenance laptops without changing network configs all the time.

The versions of logix firmware you are running will make a difference also as older versions were multicast only and some were multicast by default but you could enable unicast and the last 2-3 most recent firmware is unicast by default so it depends.

Traffic flow and loading will be determined by the amount of nodes and how the subnet and subnet mask is setup. You only provided the class info so we would need full subnet and mask schema info to give a more detailed answer.

Kev77 said:

As far as switch selection I would think about using managed switchs even if it is for the diagnostic functions.

Click to expand...

I second this.

It makes little sense to spend the kind of money that your company is spending on PLC equipment and then skimp on the Ethernet switches that are a core component of your system.

Another problem you could have with using multiple un-managed switches is if someone were to create a loop (multiple paths between the same switches), a broadcast storm could be created and take down the network. Managed switches, using some type of Spanning Tree protocol, would prevent this.

You may also have a bandwidth issue using multi-casting with un-managed switches. The IGMP snooping feature of Managed switches could help with this.

I had a lot of trouble with novel network and ab.
there is a tech note stating that rslinx and novell network is incompatable.

regards,
james

Tell them unmanaged switches with a 255.0.0.0 subnet will be fine, as long as they can deal with their PLCs losing communication and their product line shutting down.

The Plc Kid said:

IMHO each major process line should be its own network with a vlan between process lines that need to talk to each other.

This requires a layer 3 solution to be able to manage them all from the same engineering computers and maintenance laptops without changing network configs all the time.

The versions of logix firmware you are running will make a difference also as older versions were multicast only and some were multicast by default but you could enable unicast and the last 2-3 most recent firmware is unicast by default so it depends.

Traffic flow and loading will be determined by the amount of nodes and how the subnet and subnet mask is setup. You only provided the class info so we would need full subnet and mask schema info to give a more detailed answer.

Click to expand...

I included the subnet, they want to use 255.0.0.0. No gateway.

James

NFW!!!. Idiots. They are just begging for problems.

They need to get someone from rockwell in to talk to them. Sounds like your IT dept is a bunch of idiots cause any IT work his salt know better than that on any system.

By using a 255.0.0.0 subnet mask that is called supernetting and it will be pure disaster on the floor if they do this.

The system will be very unpredictable and can lose comms at any time.

Do they plan to put connect thius to the corporate network in any way? If they do they are in for a nice surprise.

The Plc Kid said:

NFW!!!. Idiots. They are just begging for problems.

They need to get someone from rockwell in to talk to them. Sounds like your IT dept is a bunch of idiots cause any IT work his salt know better than that on any system.

By using a 255.0.0.0 subnet mask that is called supernetting and it will be pure disaster on the floor if they do this.

The system will be very unpredictable and can lose comms at any time.

Click to expand...

I don't actually work for them, they are a customer. I caught wind of their plan and I'm trying to make sure they don't get burned.

This is what the IT guy told me in an email after I said I didn't recommend using a subnet of 255.0.0.0.:

"Also, if you don’t use a 255.0.0.0 subnet, your devices, mainly your software will not be able to communicate with the other devices without using some L3 switches and routing being done. Gateways will also have to be assigned to the devices."

So, it sounds like he/they have no plan on using any layer 3 switches and just think connecting everything to a "Best Buy" switch will be enough.

Thanks for all of you input. Sounds like it got some of your blood boiling.

James