Why CIP Routing is not Ethernet Networking...
No apologies necessary here...
But I do apologize if the tone of my post seemed in any way condescending. It was not intended to be. Quite the opposite in fact. Whenever I notice a statement that seems to be fundamentally incorrect, and I "think" I know better, and I have the time, then I will try to inform the person(s) as to what I believe is their misconception. I do so in an effort to teach, not ridicule...
I was trying to pose the question in a manner intended to make you stop and think about it for a while and then come back with your answer. I also added a subtle distinction into the post title to stir your mind a little. If you still came back with the "IP Conflict" type answer then that is fine too as it also informs me that my initial thinking was correct. You do not fully understand the difference between a Logix based Industrial Automation CIP Network, that happens to use Ethernet media, and any other "regular" Ethernet Network.
A design flaw perhaps?...
A fair few of us will have heard of or come across ControlNet and DeviceNet Industrial Networks.
A few of us are fairly experienced at using them.
A lesser few of us are very experienced at using them.
A lot of us will have heard of Ethernet Industrial Networks.
A lot of us "think" we are fairly experienced at using them.
A lot of us "think" we are very experienced at using them.
So much so that we will quite confidently give what we feel is experience backed advice to people that may be using what can be quite complicated Industrial Automation Control Systems (IACS), that happen to be using Ethernet media.
Huh? Why would I think that?...
Again, this is not to be condescending. The very fact that so much experience at standard Ethernet implementations can often be in place long before a user first meets Industrial Ethernet, and its implementation using CIP, that the waters are already muddied. Too much prior knowledge of the Ethernet Standards can be a dangerous thing. It's very important to grasp the whole concept of the Ethernet Standards, so as to better understand how they are implemented while using CIP. But it can often be seen, such as in this case, where a person automatically, but incorrectly, applies the Ethernet Standards in their normal way to a very different implementation to what they have experienced before.
When I first met ControlNet and DeviceNet, which also use CIP, I had to learn them from scratch. They were proprietary in every way. There were no previous systems quite like them; ODVA's Object based CIP was a new concept to Industrial Networking. Similarities could be drawn of course with others, but they were fundamentally unique in their implementations. So there was no prior knowledge of anything that could confuse you between each of them or any other networking architecture. Ethernet based CIP was not introduced until some years later. But...
I have been dealing with Ethernet Networks for as long as I can remember. Certainly long before Industrial Ethernet Networks and CIP arrived on the Allen Bradley scene. If I did not have any experience, or enough experience working with Industrial Ethernet Networks to know certain differences, some subtle, some not so subtle, then it could be quite easy for me, and of course many others, to draw upon there sometimes vast knowledge of "Ethernet" and apply it, incorrectly, to these apparently similar networks.
In short...
Many users who are quite familiar with Ethernet Networks, do not fully understand that the same rules do not always apply when dealing with Industrial Ethernet Networks.
I won't go all into ODVA and CIP and the OSI Layers. I'm off on a long enough tangent as it is. But I will put it as simple as this...
CIP is the Common Industrial Protocol Standard which was designed specifically for use in Industrial Automation Control Systems (IACS). It is an Object-Oriented Protocol which may be implemented using a number of different physical media. Therefore it is known as media-independent.
The fact that Logix based systems, such as ControlLogix, can use CIP over Ethernet media (EtherNet/IP), which happens to be the same media that many other Ethernet based protocols use, is not to be confused as these CIP systems being the same as those other systems.
Ethernet/IP, which is CIP implemented using Ethernet Standards, is an Application Layer Protocol. It defines its Transport, Network, Data Link and Physical Layers using the current Internet Protocol (IP).
(I said I wouldn't get into all the OSI Layers, etc., didn't I?...Oh well, that's blown it!)
EtherNet/IP defines the data structure to be used inside TCP or UDP data packets. For instance, using the CIP Object-Oriented model EtherNet/IP can use the well known User Datagram Protocol (UDP) to transport I/O messages, which is one of its most widely used implementations i.e. Implicit I/O Control.
In general on an Ethernet Network, Industrial or otherwise, most devices are transparent to each other on the same Subnet and so devices must be configured in a unique fashion, else there will be conflicts.
On an Industrial Ethernet Network, where many devices, or Nodes, exist on the same Subnet, they are all exposed to each other's traffic and faults. To reduce this risk you try to break down the devices into groups that only need communicate with each other. The most basic way to achieve this is by using IP Subnetting. This way the Subnet Mask will only allow devices on the same Subnet to communicate with each other.
Where you have more than one Subnet, and devices on each Subnet are configured the same, you must segregate them using a form of "Segmentation". This allows us to further separate out these individual Subnets so they can co-exist on the same wider network while using identically configured devices. A typical example is where you purchase more than one of the same machine or process that have identical networking configurations. Rather than getting into reconfiguring them all to be unique, you Segment them.
Segmentation can either be Physical or Logical. Physical means just that; you use hardware to physically create the boundaries between the network segments. An example of this is the Logix Backplane. All of the modules, whether processor, communications, etc. are physically isolated from each other via the Backplane.
Logical Segmentation may be achieved by creating Virtual Local Area Networks (VLANs). Different IP Subnetted devices can be added to the same VLAN but it is recommended to assign each IP Subnet to its own VLAN. VLANs have the added advantage of restricting the traffic to between the devices within that VLAN only. To communicate between VLANs you must use a Layer 3 Switch or Router to perform Inter-VLAN Routing.
You can assign the same IP Subnet range to multiple VLANs, as in the case of the identical machines, but you cannot use Inter-VLAN Routing between them. They must remain standalone to each other and can only be routed to other VLANs that use a different IP Subnet range. An example of this is two drives connected via Ethernet to their respective local controller. They both have the same IP address and Subnet Mask but are assigned to two different VLANs. They are both under the Implicit Control of their respective controller and do not need to directly communicate with any other device outside of their IP Subnet or VLAN. If you must route between devices on different VLANs then you must assign unique IP Subnetting. VLANs and Inter-VLAN Routing will typically be setup using Stratix 8000 series switches or their equivalent.
The more you need to break down different parts of your installation into meaningful segments, to avoid unwanted traffic and faults between unrelated sections, the more complicated the installation can appear to become. You choose whether to physically and or logically segment your Industrial Ethernet Network.
In this case here, they either by choice, or by necessity, decided to physically segment everything on the Industrial Ethernet Network in the local chassis. So the Backplane is acting as the Gateway or Router for any communications that these Ethernet modules may be passing between themselves and the local controller, or the other Ethernet modules. Even though it appears as though all the Segmentation has been carried out here in the local chassis; there may very well be heavily configured VLANs on switches connecting into this chassis. Physical and logical Segmentation.
One thing to consider is that it is entirely possible that none of them are actually talking to each other and that each one is just for segmented communications to the local controller.
The more that "you" i.e. wendaiyu, DINT2, or anyone unsure, understand about Industrial Ethernet Networks and the reasons why they may be physically and or logically segmented, the more you can appreciate how this installation may not be as convoluted as it may at first appear.
The guys have given plenty of good examples where multiple segments might be required, so I won't get into all the permutations as to why those modules might be there.
More on the Backplane to come...